text
stringlengths 3.02k
11.7k
| source
stringclasses 1
value |
|---|---|
International Journal of Information Management 60 (2021) 102383 Available online 8 July 2021 0268-4012/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons. org/licenses/by/4. 0/ ). Review Article Articial intelligence in inormation systems research: A systematic literature review and research agenda Christopher Collinsa,*, Denis Dennehya, Kieran Conboya, Patrick Mikaleb a NUI Galway, Upper Newcastle Road, Galway, Ireland b Norwegian University oScience and Technology, Høgskoleringen 1, 7491 Trondheim, Norway A R T I C L E I N F O Keywords: Articial intelligence AI Machine learning Systematic literature review Research agenda A B S T R A C T AI has received increased attentionrom the inormation systems (IS) research community in recent years. There is, however, a growing concern that research on AI could experience a lack ocumulative building oknowledge, which has overshadowed IS research previously. This study addresses this concern, by conducting a systematic literature review oAI research in IS between 2005 and 2020. The search strategy resulted in 1877 studies, o which 98 were identied as primary studies and a synthesise okey themes that are pertinent to this study is presented. In doing so, this study makes important contributions, namely (i) an identication othe current reported business value and contributions oAI, (ii) research and practical implications on the use oAI and (iii) opportunitiesoruture AI research in theorm oa research agenda. 1. Introduction AI has been claimed to oer transormational potential across sectors and industries, rangingrom supply chain management ( Chi, Huang,& George, 2020; Collins, Youngdahl, Jamison, Mobasher,&Gini, 1998; Nissen&Sengupta, 2006; Rodriguez-Aguilar, Martin, Noriega, Garcia,& Sierra, 1998 ) to medicine ( Ali, Shrestha, Soar,&Wamba, 2018; Cepo-lina&Muscolo, 2014; Mettler, Sprenger,&Winter, 2017; Wang, Savkin, Clout,&Nguyen, 2015 ) to automobiles ( Lugano, 2017 ). Studies have reported that AI provides opportunities to reinvent business models (Duan, Edwards,&Dwivedi, 2019 ), change theuture owork ( Schwartz et al., 2019 ), perormance improvements ( Wilson&Daugherty, 2018 ), and even enhance human capabilities ( Dwivedi, et al., 2021 ). The heightened interest in AI to transorm economies ( Majchrzak, Markus,&Wareham, 2016; Ransbotham, Fichman, Gopal,&Gupta, 2016; Watson, 2017 ) is refected in the scale oglobal spending, which the International Data Corporation (IDC) predicts that global spending on AI will reach nearly $98 billion in 2023, more than twice the $37. 5 billion that was spent in 2019. Yet, there is no consensus on what denes AI or what distinguishes itrom other digital technologies ( Bhatnagar et al., 2018; Monett&Lewis, 2018; Nilsson, 2009 ). There are aew reasons attributed to this upswing in AI interest in recent years ( von Krogh, 2018 ). The pastew decades have seentremendous advancements in some othe underlying AI methods such as current and convential neural networks, many owhich have been made open-source and thus available to everyone. AI requires extensive and sophisticated computation, so the decreasing cost in computer hardware and dedicated AI chip designs is making it much moreeasible and thus attractive to organisations. The expansion ocloud-based services related to AI has also made it much more attainableor organisations would otherwise be hesitant. Additionally, while research is still in the early stages, preliminary studies may be showing that the advent o COVID has also increased interest in AI, as people get used to the reduced human element in all levels osociety and the increased use o automation ( Coombs, 2020; Sipior, 2020 ). The aim othis research is to understand the various characteristics oAI studied within the context oIS. A systematic literature review is important as it can be used to provide a valuable baseline to aid in urther research eorts ( Kitchenham, Budgen,&Brereton, 2011; Petersen, Vakkalanka,&Kuzniarz, 2015 ). The aims othis systematic review are to: 1. identiy the reported business value and contributions oAI 2. examine the practical implications on the use oAI 3. identiy the opportunitiesoruture AI research in IS. * Corresponding author. E-mail addresses: C. Collins20@nuigalway. ie (C. Collins), denis. dennehy@nuigalway. ie (D. Dennehy), kieran. conboy@nuigalway. ie (K. Conboy), patrick. mikale@ntnu. no (P. Mikale). Contents lists available at Science Direct International Journal oInormation Management journal homepage: www. elsevier. com/locate/ijinfomgt https://doi. org/10. 1016/j. ijinomgt. 2021. 102383 Received 5 November 2020; Received in revisedorm 21 June 2021; Accepted 22 June 2021 | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 2The structure othe paper is asollows. First, an introduction to related work on AI in the ISeld is presented. Then the methodology o the systematic literature review is explained, and limitations othe study are acknowledged. Next, state-o-the-art oAI research is presented, includes the reported business value and contributions oAI, and anal-ysis on how AI is dened. Followed by a discussion, implications, and a research agendaor theuture. The paper ends with a conclusion and directionsoruture research. 2. Background and related work This section commences with an overview oextant inormation systems literature on AI. The lack oclarity concerning the concept and classication oAI are discussed. 2. 1. Evolution oAI defnition AI has a history much longer than is commonly understood, inelds rom science and philosophy ranging all the way back to ancient Greece (Dennehy, 2020 ), but its modern iteration owes much to Alan Turing (Turing, 1950 ) and conerence in Dartmouth College in 1956 ( Mc Cor-duck, 2004 ), where the term“Articial Intelligence” was ocially coined and dened by John Mc Carthy at the time as“the science and engineering omaking intelligent machines”. Russel and Norvig (2020) reerred to it as the“the birth oarticial intelligence. ” One othe initial paradigms oAI was that it revolved around high-level cognition. Not the ability to recognise concepts, perceive objects, or execute complex motor skills shared by most animals, but the potential to engage in multi-step reasoning, to understand the meaning onatural language, to design innovative arteacts, to generate novel plans that achieve goals, and even to reason about their own reasoning ( Langley, 2011 ). This general human like intelligence was reerred to as strong AI ( Kurzweil, 2005 ). For strong AI, the primary approach has centred on symbolic reasoning, that computers are not simply numeric calculators but rather general symbol manipulators. As noted by Newell and Simon (1976) in their physical symbol system hypothesis, intelligent behaviour appears to require the ability to interpret and manipulate symbolic structures. While this approach showed promise initially ( Newell&Simon, 1963 ), many branches oAI have retreatedrom this approach due its diculty and the lack oprogress coming in to the 21stcentury. It remains yet uncertain on when and istrong AI will be made a reality. The distinction between weak AI and strong AI is also concerned with rule adherence, i. e., the way machines interact with rules. Wole (1991) distinguishes rule-based decision making in which machines strictly respect the rules set by developersrom ruleollowing decision making which machinesollow rules that have not been strictly specied to them. Rule-based decision-making matches weak AI, while rule-ollowing decision making is an attempt that tends towards strong AI. An example orule-ollowing decision making is neural networks (NN), which allow algorithms to learnrom themselves. Strong AI would be machines making their own rules and thenollow them, which is not possible at the stage oright now ( Wole, 1991 ). AI has gone through many peaks and troughs since its early inception in the 1950s, usually reerred to as AI“summers and winters” ( Russel and Norvig, 2020 ). Since 2010, however, AI can be said to have once again entered a summer period, mainly due to considerable improve-ments in the computing power ocomputers and the access to massive amounts odata ( PWC, 2019 ). This resurgence in AI research is the result othree breakthroughs: (1) the introduction oa much more sophisti-cated class oalgorithms; (2) the arrival on the market olow-cost graphics processors capable operorming large amounts ocalcula-tions in aew milliseconds; and (3) the availability overy large, correctly annotated databases allowingor more sophisticated learning ointelligent systems ( Jain, Ross,&Prabhakar, 2004; Khashman, 2009; PWC, 2019 ). Despite the length otime theeld has existed, there is still nocommonly accepted denition ( Allen, 1998; Bhatnagar et al., 2018; Brachman, 2006; Hearst&Hirsh, 2000; Nilsson, 2009 ). This is not considered a problem yet, as many scientic concepts only get true denitions ater they have matured enough, rather than at their conception, and given the complexity and breadth oAI, it may not be easible to expect AI to have a set denition yet. Still, this doesn't mean that the topic should be ignored, especially with the recent advance-ments and advancements relating to theeld ( Le Cun, Bengio,&Hinton, 2015; Silver et al., 2016 ). However, without a clear denition othe term,“it is difcultor policymakers to assess what AI systems will be able to do in the nearuture, and how the feld may get there. There is no common ramework to determine which kinds oAI systems are even desirable” (Bhatnagar et al., 2018 ). A similar concern has been echoed by Monett and Lewis (2018), that“theories ointelligence and the goal oArtifcial Intelligence (A. I. ) have been the source omuch conusion both within the feld and among the general public”. In the years immediately preceding and ater the 1956 Dartmouth conerence where the term was coined, when the conceptor AI wasrst brewing in academic consciousness, many researchers (would later becomeamousor their contributions to AI)ormulated many theories and proposals thatocused on the commoneatures omind and (Mc Culloch&Pitts, 1943; Turing, 1950; von Neumann, 1958; Wiener, 1948 ). While these thought leaders were infuential, theeld oAI as we know it owes more to Mc Carthy, Minsky, Newell, and Simon. While this is partly due to their own attendance otheamous 1951 Dartmouth conerence, it is likely more since they went on to establish three leading research centres which shaped the stream othough regarding AIor years. Their own opinion on AI was asollows; “By 'general intelligent action' we wish to indicate the same scope o intelligence as we see in human action: that in any real situation behav-iour appropriate to the ends othe system and adaptive to the demands o the environment can occur, within some limits ospeed and complexity” (Newell&Simon, Computer science as empirical enquiry: Symbols and search, 1976). Intelligence usually means“the ability to solve hard problems” ( Min-sky, 1958 ). “AI is concerned with methods oachieving goals in situations in which the inormation available has a certain complex character. The methods that have to be used are related to the problem presented by the situation and are similar whether the problem solver is human, a Martian, or a computer program” ( Mc Carthy, 1988 ). With the variety oseparate opinions on what AI is, lacking agree-ment on a standard evaluation (i. e., criteria, benchmark tests, mile-stones) makes it extremely challengingor theeld to maintain healthy growth ( Hern´andez-Orallo, 2017 ). 2. 2. Previous systematic literature reviews oAI in IS research Despite the heightened interest in AI ( Watson, 2017 ), it is claimed that there is a noticeable absence otheoretically-grounded research on how organisations should develop their digital business strategies incorporating AI to create business ( Mikale, Bourab, Lekakosb,& Krogstiea, 2019 ). We investigate this claim and identiy gaps in AI research conducted by IS scholars. We acknowledgeour previous Sys-tematic Literature Reviews (SLRs) have been conducted ( Homann, Oesterle, Rust,&Urbach, 2019; Rzpeka&Berger, 2018 ;Borges, Laur-indo, Spínola, Gonçalves,&Mattos, 2021 ;Karger, 2020 ) but highlight limitations othese studies (see Table 1 ). The literature review conducted by Rzpeka and Berger (2018) cited 91 studies total as their primary studies, takenrom a combination o conerences and journals. However, it isocused on the context oin-dividual user interaction with AI systems in IS, while this study studies how it is being dened and gives value. The literature review conducted C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 3by Homann et al. (2019) was primarily concerned with the eects oAI and ML in the context othe radiology value chain, and so had a time span refecting that,ocusing on 2012 to 2018. The review conducted by Borges et al. (2021) ocused more on specic AI interactions with organisational strategy and so misses some othe context in how it is being dened and how it creates value. The literature review conducted by Karger (2020) isocused narrowly on the relations between AI and blockchain and excludes everything else. This study includes all the relevant studies in ateen year period in a number ohigh quality journals and conerences, and includes studies that use AI in more oblique ways than these SLRs, such as studies that use machine learning approaches when researching theirocus. 2. 3. AI Functions The current diculty to settle on an agreed denition oAI has been discussed above, butor the purposes othis systematic literature re-view, weocus onunctions oAI as described by Dejoux and L ´eon (2018). The broad range o unctions is shown in Table 2 below. Already, studies are showing the potential opportunities oadopting AI in a wide range o elds, with manuacturing, digital marketing and healthcare generating considerable academic interest ( Juniper Research, 2018 ). For manuacturing,actories are likely to extensively use AI as product automation increases and industry uses more AI and cyber physical systems ( Wang&Wang, 2016 ). Healthcare researchers propose using AI systems linked to sensors placed on humans to monitor and record their health ( Rubik&Jabs, 2018 ). For digital marketing, Juniper Research (2018) predicts that demandorecasting using AI will more than treble between 2019 and 2023 and that chatbot interactions will reach 22 billion in the same yearrom current levels o2. 6 billion. However, these opportunities are only available ione can undertsand what AI is. 3. Research methodology This section outlines the systematic review process adoptedor this study. A Systematic Literature Review is dened as“means oidenti-ying, evaluating and interpreting all available research relevant to a particular research question, or topic area, or phenomenon ointerest” (Kitchenham, 2004 ). This systematic approach was chosenor its ability to oer reviews ohigh quality ( Dingsøyr&Dybå, 2008 ) and transparent and replicable review ( Leidner&Kayworth, 2006 ). Additionally, it is useulor studies with a clearlyormulated research question ( Petticrew &Roberts, 2006 ) and summarising large quantities oresearch studies (Fink, 2005 ). Thus, the SLR was chosenor theollowing reasons: (i) the study will generate large amounts oliterature; (ii) this study aims to answer a specic research question; (iii) we intend to systematically extract relevant AI reerencesrom the studies transparently; and (iv) the rigour and replicability it oers leads to an unbiased scientic study. Theoundation oour guide was takenrom the guideline developed by Okoli (2015). 3. 1. A systematic guide to literature review development Okoli (2015) propose a systematic review process that consists o8 steps, namely planning (2 steps), selection (2 steps), extraction (2 steps) and execution (2 steps) that are completed across 4 phases (see Fig. 1. ). Each otheseour phases and eight steps are discussed in detail in the remainder othe section. The objective othe systematic literature review is to answer the research questions shown in Table 3. RQ2 is a comprehensive research question. Five questions (RQ2. 1-RQ2. 5) are used to answer RQ2. RQ1Table 1 Comparison oprevious SLRs in IS. Comparison Element Purpose Years Included Number o primary studies Rzpeka and Berger (2018)Provides insight into individual user interaction with AI systems in IS. 1987-2017 91 Homann et al. (2019)Identies opportunities and challenges oML across the radiology value chain. 2012-2018 29 Borges et al. (2021)Studies the integration o AI and organisational strategy2009-2020 41 Karger (2020) Studies the interactions between Blockchain and AINot specied (primary studies rangedrom 2014 to 2020)32 This Study AI, as a subject and as a use in theeld oIS2005-2020 98 Table 2 AIunctions. Title ( Dejoux& L´eon, 2018, p. 188)Description ( Brynjolsson& Mc Aee, 2014 ;Jarrahi, 2018a, 2018b )Example ( Dejoux&L ´eon, 2018, p. 188) Expert Systems (ES)Designed to simulate the problem-solving behaviour o a human. DENDRAL: Expert system used or chemical analysis to predict molecular structure. Machine learning Automatically renes its methods and improve its results as it gets more data. Many othe more advanced recommendation systems i. e., Google, You Tube etc. Robotics Concerned with the generation ocomputer-controlled motions o physical objects in a wide variety osettings Service robots Natural Language Processing (NLP)Designed to understand and analyse language as used by humans. NLP is the baseor the AI-powered Speech Recognition. Intelligent agents i. e., Apples Siri, Amazons Alexa Machine vision The analysis oimages using algorithmic inspection. The computer vision used to help drive autonomous vehicles Speech recognition Can be understood as an approach that deals with the translation ospoken words into the text. Google Dictate uses speech recognition to convert spoken words into text Fig. 1. A systematic guide to literature review development ( Okoli, 2015 ). C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 4and RQ3 were used to create rich dataor the synthesis and discussion stages. Additionally, this review also aims to contribute to conducting an IS SLR. Aurther contribution will be that oa study which conducts a SLR: (i) where the complexity and type oAI is incorporated into a search strategy; (ii) tond relevant AI studies; (iii) which are then systematically analysed. A literature review's quality is dependent on the rigor othe search process ( Vom Brocke et al., 2009 ). Thereore, the search strategy is best developed in concert with the research question. The goal is tond as many studies as possible capable oanswering the research question (Kitchenham&Charters, 2007 ). The starting point when searching literature is electronic sources and literature databases,ollowed by a keyword string search to locate the appropriate studies ( Levy&Ellis, 2006 ). A generally accepted approach to search string strategy is to base the search string on the research questions and include a list osynonyms, abbreviations, and alternative spellings ( Kitchenham&Charters, 2007 ). Due to the nature oArticial Intelligence and the variety odierent types, subtypes, and methods in use, in addition to the dierent ways it is reerred to by researchers, a thorough strategy was needed. The search string was usedollowing the Boolean practice. A simple“OR” operator was used between keywords. The use o“*” ater some word was implemented so the search would include multiple variations othe word. The use oquotation marks (“”) over some terms was to exclu-sively searchor that specic term. The selected databases are pertinent to this study as these return the most studies ( Dingsøyr&Dybå, 2008 ). For each othe three selected databases (AIS e Library, Scopus, ISI Web oScience), using the specied search string retrieves an initial list ostudies. One or many othese databases had been used by multiple researchers in the literature ( Agarwal, Kumar,&Goel, 2019; Busalim& Hussin, 2016; Gupta, Kar, Baabdullah,&Al-Khowaiter, 2018; Rekik, Kallel, Casillas,&Alimi, 2018 ) The records arerst imported into Endnoteor sorting and categorisation, and thenurther imported into Microsot Excel sheetormat. The basic input includes meta-data such as (i) title, (ii) author, (iii) year, (iv) publication type, and (v) abstract. The keyword strategy was applied to nine IS journals and the top two IS conerences. The search string used across the three databases retrieved 1877 studies (shown in Table 4 ). Once the literature searches in leading journals have commenced, the leading conerences oaeld shouldollow ( Webster&Watson, 2002 ). Due to the multidisciplinary nature oIS literature is heavily dispersed across dierent data sources ( Levy&Ellis, 2006 ). To ensure all relevant studies were retrieved, three databases were used, these being (i) AIS e Library; (ii) Scopus, and (iii) the ISI Web oScience. AIS e Library was used as othe databases used it was unique in being able to provide studiesrom the leading IS conerences. Scopuswas used as it claims to be largest databaseor abstracts and citations ( Ballew, 2009; Kitchen-ham&Charters, 2007 ). ISI Web oSciencewas used as it is the largest citation database which stores over 800 million reerences ( Manikandan&Amsaveni, 2016 ). These databases were used to retrieve studiesrom relevant IS journals, which were chosen as we beliethese nine journalsocus on the social-technological aspects oAI, which is the scope othis SLR. i. International Journal oInormation Management ii. Management Inormation Systems Quarterly iii. Journal othe Association oInormation Systems iv. Inormation Systems Journal v. Inormation Systems Research vi. Journal oInormation Technology vii. Journal oManagement Inormation Systems viii. Journal oStrategic Inormation Systems ix. European Journal oInormation Systems Conerence papersrom two othe leading IS conerences, namely, International Conerence on Inormation Systems(ICIS) and the European Conerence on Inormation Systems(ECIS) were also retrieved and ana-lysed. The number opapers retrievedrom each selected database is shown in Table 5. Screening othe retrieved studies was achieved byollowing the best practices proposed by Kitchenham (2004) and Dingsøyr and Dybå (2008). The study selection process used in this study is illustrated in Fig. 1. Two authors independently analysed the 1877 studies to remove (i) duplicate studies, (ii) non-English studies, (iii) non-IS studies, and (iv) non-peer reviewed scientic studies (books, book chapters, experi-ence reports). As searching through literature can result in many studies, using an inclusion and exclusion criteria can serve to eliminate unnec-essary studies ( Okoli, 2015 ). Studies were eligibleor inclusion in the systematic review ithey presented empirical data on AI or ML in IS, or oAI and ML being used in the IS literature, or ia non-empirical study shows clear evidence o academic rigour. The inclusion criteria applied was; The studies should be written in English. Table 3 Research questions. ID Research question RQ 1 How is AI being dened in theeld oIS? RQ 2 What is the current state oAI in IS? RQ 2. 1What number oIS academic studies on AI has been published between 2005 and 2020? RQ 2. 2What were the Publication channels used? RQ 2. 3What were the research methods and data collection techniques used? RQ 2. 4What kind ocontributions are provided by studies on AI in IS? RQ 2. 5What AIunctions are used by IS researchers? RQ 3 What is the business value oAI?Table 4 Search string. Source String Science Direct-IJIM AIS e Library Web oScience("AI" OR“articial intelligence” OR "machine learning" OR "neural networks" OR cognitive* OR automation* OR robot* OR augment*) SCOPUS TITLE-ABS-KEY ("AI" OR“articial intelligence” OR "machine learning" OR "neural networks" OR cognitive* OR automation* OR robot* OR augment*) AND SRCTITLE ("MIS Quarterly: Management Inormation Systems" OR "INFORMATION SYSTEMS RESEARCH" OR "JOURNAL OF MANAGEMENT INFORMATION SYSTEMS" OR "JOURNAL OF STRATEGIC INFORMATION SYSTEMS" OR "EUROPEAN JOURNAL OF INFORMATION SYSTEMS" OR "INFORMATION SYSTEMS JOURNAL" OR "JOURNAL OF INFORMATION TECHNOLOGY" OR "JOURNAL OF THE ASSOCIATION FOR INFORMATION SYSTEMS") AND PUBYEAR=2020 Table 5 Selected databases and retrieved papers. Database Filter No. oretrieved studies AIS e Library Only conerence papers 468 Scopus Only conerence papers and journal articles399 ISI Web oScience Only conerence papers and journal articles210 International Journal o Inormation Management Only conerence papers and journal articles800C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 5The studies should be published between 2005 and 2020. The studies directly answer one or more othe research questions o this study. The studies should clearly state itsocus on AI/ML or use AI/ML as a large part otheir methodology. For example, publications that explicitly use a machine learning approach as aundamental part o their methodology/research. Ithe studies have been published in more than one journal or con-erence, the most recent version ostudies is included. Opinion/perspective studies were included as we believe that ithey were published in the relevant journals than they could be used to gain insight into the state oAI in IS. The exclusion criteria applied was; Not written in English. Duplicate articles. Simulation studies. Editorials. Studies with noocus on AI. Non-peer-reviewed scientic publications (editorials books, book chapters, articles). Fig. 2 shows the study selection process. Ater the initial step o identiying a search string was complete, pilot steps were carried out on the databases used. This entailed rening the search stringor each database. However, the terms used to search the databases were used the same throughout. The lead author analysed the 1877 studies retrieved in the initial search. A second reviewer was invited to analyse these studies as well, to pre-emptively combat potential bias. The two reviewers had to agreeor a study to stay a primary study. Based on the removal o duplicates, non-scientic and non-English studies, 91 were removed, leaving 1786. These 1786 studies were then analysed based on title. The title gave a clear indication on whether they were outside theocus o the study, and thus excluded. Ia title did not clearly reveal application domain othe study it was includedor review in the subsequent steps, where title, keywords and abstract were examined. Based on title, ab-stract and keyword, the 1786 studies wereurther narrowed down to 187. There were still cases where the abstract was unclear, so thesestudies carried onto the next stage, where the contents otheull study were examined. An in-depth examination othe 187 remaining studies was undertook by the reviewers, which resulted in aurther 90 studies being excluded. This resulted in a total o98 primary studies used as the basis othis SLR. Thendings and analysis othese 98 primary studies is presented in the next section. 3. 2. Threats to validity and limitations ostudy There are always several common threats to validity concerning SLRs (Petersen, Vakkalanka,&Kuzniarz, 2015; Wohlin et al., 2012 ). This section considers those threats and outlines the strategies used to com-bat and mitigate them, as well as explores the limitations othis study. The validityramework by Wohlin et al. (2012) examines validity threats in terms o(i) construct validity, (ii) external validity, (iii) in-ternal validity, and (iv) conclusion validity. Construct validitystates that the author must attain the right measuresor the concept under study (Petersen, Vakkalanka,&Kuzniarz, 2015; Wohlin, et al., 2012 ). To minimise this threat, this SLRollowed a structured eight-step guideline required to conduct a scientically rigorous SLR, as seen in Fig. 1. Within those guidelines was the paper selection process (see Fig. 2. ) which documents the process o ltering studiesrom the original 1877 to the 98 primary studies. Tourther mitigate this threat, authors three andour were experienced in reviewing studies and acted as external reviewers to validate the research protocol. This, this threat has been signicantly neutralised. External validityisocused on the generalisability othe study. That is, the extent to which the study can be generalised to other areas outside othe context othis study ( Petersen et al., 2015; Wohlin et al., 2012 ). To know to what degree the results oa study can be generalised it is essential that the research process is described ( Petersen&Wohlin, 2009 ). As this systematic studyollowed the eight-step guideline laid out by (Okoli, 2015 ), it is attributed to mitigating the threats to validity. Internal validityrelates to causal relationships and ensuring that it is not a result oaactor that was not measured, or the researcher had no control over. As the aims othis study were not to establish a statistical causal relationship on AI in IS, other mitigations were used to combat it, such as regular meetings with all authors to explore any potential o bias. Conclusion validityrelates to bias othe researchers in the inter-pretation othat data. While this risk cannot be eliminated, several measures were taken to combat it; (i) three authors were involved in data extraction othe primary studies, (ii) aull 'audit trail' rom the initial 1877 studies to the identication o98 primary studies was provided, and (iii) conclusions drawnrom analysis othe 98 primary studies involved all authors. Although this paper concentrated on mitigating threats to validity using well-established strategies, we acknowledge that publication bias is a limitation othis study, as weocused on a select number oIS journals, meaning that other studiesrom IS conerences and non-IS outlets were excluded. 4. Findings and analysis This section presents the resultsrom the analysis othe 98 primary studies, based on the research questions listed previously. The results represent the state oAI research in IS and is based on theollowing (i) how AI is being dened, (ii) study by year, (iii) publication channel, (iv) research methods adopted, (v) type ocontribution, (vi) types oAI and (vii) the reported business value oAI. 4. 1. RQ1: how is AI being defned in the feld oIS? The aim othis research question is to identiy and analyse the dierent denitions oAI used in theeld oIS. It was noted in Section 2. 1 the diculties theeld oAI had with denitions, and this research Fig. 2. Study selection process. C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 6questions aims to look at how IS has handled those diculties. However, despite AI and Machine Learning being a large part othe primary studies, many did not provide a denition, or used denitions that were not cited (see Fig. 3 ). Othe 98 primary studies, 54 othem gave no clear denition othe AI relevant to the study. And othe 44 studies that did give a denition, 7 othem cited no reerenceor the denition given. The denitions o AI used in the primary studies varied in both term odenition and source cited. Disregarding the seven studies that dened AI without citing a source, Russel&Norvig's book Artifcial Intelligence: A Modern Approachwas the mostrequently cited sourceor dening AI, though the actual edition othe book varied, with each studies using the latest edition at the time. The denitions and cited sources othe primary studies can be seen in Table 6. 4. 2. RQ2: what is the current state oAI in IS? This aim othis research question is to examine the current state oAI in theeld oIS through a series osub-related research questions. 4. 2. 1. RQ 2. 1. What number oIS academic studies on AI has been published between 2005 and 2020? The aim othis research question is to identiy the number oaca-demic studies involving Articial Intelligence and Machine Learning in theeld oInormation Systems, specically those between the years 2005 and 2020 (see Fig. 4. ). Fig. 4. reveals that studies on AI remained relatively lowor most othis period, with a total o11 studies between the years 2005 and 2015. 2019 and 2020 show an immense surge in AI related studies in IS, signiying a much greater interest in theeld. Due to the inclusion and exclusion criteria othis study, there were no studies on AI in Inormation Systems in the years 2007, 2008, 2010, and 2012. 4. 2. 2. RQ 2. 2 What were the publication channels used? The aim othis research question is to identiy the main channels where AI studies are disseminated. Table 7 shows that 32 othe primary studies were published in journals and 65 were published in the top two IS conerences. The highest number ostudies were published in ICIS, a total o41 studies over the 15-year period. The journal with the most studies was IJIM with 14 studies, especially notable as the scopeor IJIM wasve years in comparison to the 15 years othe other journals. 4. 2. 3. RQ 2. 3 research methods and data collection techniques used The aim othis research question is to identiy the research methods and data collection techniques which were used to study AI and Machine Learning in the ISeld. Each study was either empirical, theoretical, conceptual, or experimental. Analysis was conducted to determine the Fig. 3. AI denition cited. Table 6 AI denitions in primary studies. Source Cited Defnition Primary studies source N/a P6, P21, P38, P39, P59, P61, P63 Bush (1945) A kind odeep learning machine that has the ability to create intelligent agents based on the concepts, positions and patterns oargument. P7 Mc Carthy (1958) Reerred to as“the science and engineering omaking intelligent machines” P68 Carbonell et al. (1983) ML is based on inductive learning and iner general conceptsrom example data, rangingrom simple memorising o acts that doesn't requires any inerence at all (rote learning) over learning perormed by instruction, by analogy, and by examples to learning by observation with increasing need oinerence P36 Trappl (1986) Articial intelligence (AI) was dened as: 1) making computers smart, 2) making models ohuman intelligence, and 3) building machines that simulate human intelligent behaviour P24 Russel and Norvig (1995)activities that we associate with human thinking, activities such as decision-making, problem solving, learning” P47 Jennings et al. (1998) A computer system, situated in some environment that is capable ofexible autonomous action in order to meet its design objectives. P12 Cross (2003) Dened as an Inormation Technology (IT) programs that perorm tasks on the user's behalindependently o direct control othe users themselves. P35 Russel and Norvig (2010)Articial intelligence (AI) enables the machine to exhibit human intelligence, including the ability to perceive, reason, learn, and interact, etc. P31, P44, P50 Min (2010) AI concerns understanding and learning the phenomena ohuman intelligence and to design computer systems that can imitate human behavioural patterns and create knowledge relevant to problem-solving P56 Austin et al. (2013) ML is an exploratory process where the accuracy and perormance o models vary, based on the characteristics ovariables and observations in a study P92 Deng and Yu (2014) The main idea odeep learning consists omultiple layers o eatures representations at increasing levels o abstraction P48 Le Cun et al. (2015) A set omultiple interconnected layers oneurons, inspired by the human brain, which can be trained to represent data at high levels o abstraction P53, P55 International Federation o Robotics (2016)Under the label service robots, these types odevices are designed to“[... ] operate semi-orully autonomously to perorm services useul to the well-being ohumans [... ]” P11 Goodellow et al. (2016)Deep learning is the state-o-the-art machine learning method that builds upon large-scale neural networks and unsupervised representation learning P19, P94 De Canio (2016) AI is the broad suite otechnologies that can match or surpass human P69, P97 (continued on next page)C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 7research methods and data collection techniques othe 98 primary studies. Othe 98 primary studies, eight adopted a mixed method approach, thirteen took theorm oa literature review,orty-our used a quantitative method and thirty-one used the qualitative method. The method and technique adoptedor each othe primary studies is listed in Table 8. A deeper analysis othe research methods was conducted to establish the data gathering techniques used in the primary studies. These are listed in Table 9. Many othe studies used multiple techniques to gather data in their studies, so some reerences in the table below are repeated. 22 studies used data mining and 21 reported the use oexperiments. 10 studies adopted observation techniques andour collected data through documentation. Surveys were incorporated by 8 primary publications to collect data, with interviews being adopted by 20. Literary analysis was used by 22 othe studies, and questionnaires incorporated by two. Four studies adoptedocus groups to collect data, with three using work-shops. Sample analysis was used by two studies. Theory-as-discourse, Machine Learning subset selection, causal mapping and prediction markets were all used by a single study each. Experiments were the most popular choiceor collecting data. This usually took theorm othe studies putting their relevant use oAI or Machine Learning into practise as an experiment and collecting data rom the result. As AI is still a broad and relatively nebulouseld in IS, perorming experiments can provide rich data on AI usage in a variety o complex, contextual environments. 4. 2. 4. RQ 2. 4. What kinds ocontributions are provided by studies on AI in IS? The aim othis research question is to identiy and categorise the contributions othe primary studies. These contributions (see Table 10 ), adaptedrom Shaw (2003) and Paternoster, Giardino, Unterkalm-steiner, Gorschek, and Abrahamsson (2014) include six types ocon-tributions, namely (i)ramework, method, technique, (ii) guidelines, (iii) lessons learned, (iv) model, (v) tool, and (vi) advice/implication. The contributions othe 98 primary studies and the data collection techniques that led to these contributions are listed in Table 11. This would provide signicant practical contributions, as well as widening the academic discourse on AI in IS. Analysis othe 98 primary studies also shows that contributions were largely made as 'lessons learned' (39 studies), 'methods' (26 studies), 'advice or implication' (15 studies), guidelines (6 studies), tools (5 studies) and models (6 studies). Table 11 highlights the needor research to contribute to the categories o(i) models and (ii) tools. Although several primary studies could potentially contribute to more than type one contribution type, the categorisation used in this systematic literature review is based on the primary contribution as stated by the authors oeach othe 98 primary studies. A visual repre-sentation othe contribution types othe primary studies is shown in Fig. 5. Fig. 5. shows that 'lessons learned' (40 studies) and 'methods' (26 studies) are the most popular contribution oAI studies in IS research. A limitation to these contributions is that they can oten be context spe-cic, especiallyor the methods contributions. This is compounded upon by AI encompassing such a broad array ouses and designs. This means there is less repetition ocontributions and lessons learned, and that cumulative building oknowledge in this context may take time to accumulate. For example, a study that proposes a newrameworkor improving the radiology supply chain may not be aligned with a study that examines how AI can improve drug innovation. Table 6(continued) Source Cited Defnition Primary studies source capabilities, particularly those involving cognition such as learning and problem solving Hengstler et al. (2016) Intelligent automation is reaching a level, where it is capable o perorming complex tasks that normally involve human experience and intuition P82 Russel and Norvig (2016)dened the term AI to describe systems that mimic cognitive unctions generally associated with human attributes such as learning, speech and problem solving. P58, P76, P77, P79 Günther et al. (2017) The term algorithmic intelligence reers to business analytics applications and to articial intelligence (AI) in domains such as robotics, machine vision, natural language processing, expert decision-making, and classication P10 Li et al. (2017) AI is the general conceptor computer systems able to perorm tasks that usually need natural human intelligence, whether rule-based or not, while ML is that subset oAI that is capable o“learningrom data and making predictions and/or decisions” without human dictated rules. P14 Kolbjørnsrud et al. (2017)Articial intelligence is dened as a subset oIT that can sense their environment, comprehend the collected inormation, learn, and derive actions based on interpreted inormation and their implemented objectives. P70 Nichols (2018) Robot can be dened as a programmable machine which is capable osensing and manipulating its surroundings while perorming complex tasks semi/ully autonomously P66 Plastino and Purdy (2018)Articial Intelligence is a specialorm oan IT resource with hybrideatures oan IT arteact and human capital P78 Sutton and Barto (2018)AI agents learn by themselves to achieve the optimal strategies by sequentially interacting with environments in a trial-and-error way only with the supervision orewards or punishments P81 von Krogh (2018) AI can broadly be described as a collection ocomputer-assisted systems able to perorm non-trivial tasks traditionally conned to humans P88 Rai et al. (2019) Articial Intelligence (A. I. ) is dened as the ability oa machine to perorm cognitiveunctions that we associate with human minds, such as perceiving, reasoning, learning, interacting with the environment, problem solving, decision-making, and even demonstrating creativity. P16, P87 Duan et al. (2019) Articial intelligence (AI), in which machines can“learnrom experience, adjust to new inputs, and perorm human-like tasks” P60 Berente et al. (2019) Dene AI as machines perorming the cognitiveunctions typically associated with humans, including perceiving, reasoning, learning, interacting, etc P75 Longoni et al. (2019) We dene articial intelligence (AI) as algorithms that perorm perceptual, cognitive, and P84Table 6(continued) Source Cited Defnition Primary studies source conversationalunctions typical othe human mind C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 84. 2. 5. RQ 2. 5. What types oAI technologies are used by IS researchers? The aim othis research question is to categorise studies on AI based on the type oAI used in the primary study. It is possible to combine a variety othese types together into a single AI system. For example, IBM's Watson combines NLP, ML and machine vision techniques ( Jar-rahi, 2018a, 2018b ). However,or the purpose othis SLR, the study will be categorised solely based on the primary AI type othe study (see Fig. 6. ). Fig. 6 shows that ML is the most popularorm oAI used in the Fig. 4. Number ostudies by year. Table 7 Primary studies by journal and conerence. Channel Title Number o studies Primary studies Journal (n=9)MIS Quarterly Management Inormation Systems3 P3, P4, P12 Inormation Systems Research1 P1 Journal o Management Inormation Systems3 P2, P13, P94 European Journal o Inormation Systems2 P7, P11 Journal oStrategic Inormation Systems3 P10, P93, P97 Journal o Inormation Technology3 P6, P8, P95 Journal oThe Association o Inormation Systems3 P5, P9, P96 Inormation systems journal0 International Journal oInormation Management14 P56, P57, P58, P59, P60, P61, P62, P63, P64, P65, P66, 67, P68, P69, P98 Conerence (n=2)International Conerence o Inormation Systems41 P14, P16, P17, P18, P19, P21, P22, P23, P24, P26, P27, P30, P33, P38, P39, P40, P42, P43, P44, P46, P47, P49, P50, P51, P52, P54, P55, P80, P81, P82, P83, P84, P85, P86, P87, P88, P89, P90, P91, P92 European Journal o Inormation Systems24 P15, P20, P25, P28, P29, P31, P32, P34, P35, P36, P37, P41, P45, P53, P70, P71, P72, P73, P74, P75, P76, P77, P78, P79 Total 98 98Table 8 Research method and technique adopted. Method Technique Pr imary studies Total Quantitative Survey Descriptive Case Study Deep Learning P1, P3, P4, P9, P13, P16, P18, P19, P21, P23, P27, P29, P33, P34, P35, P38, P39, P40, P42, P46, P48, P51, P52, P53, P55, P57, P59, P62, P63, P64, P66, P70, P71, P81, P82, P84, P86, P87, P89, P90, P91, P94, P95, P9844 Qualitative Interview Case Study Design Science P2, P7, P8, P12, P14, P17, P20, P22, P26, P28, P32, P41, P43, P47, P49, P50, P56, P61, P67, P69, P73, P74, P75, P76, P77, P79, P80, P85, P88, P93, P9631 Literature Review Systematic Monographic P5, P6, P10, P15, P25, P36, P37, P44, P45, P60, P68, P83, P9713 Mixed Method Survey Focus group Q-Methodology P11, P30, P31, P54, P58, P72, P78, P928 Action Research Canonical Action Research P24 1 Table 9 Data collection techniques. Data collection technique Primary studies Data Mining (n=22) P1, P3, P13, P33, P40, P46, P48, P51, P52, P53, P59, P62, P66, P74, P78, P81, P85, P89, P92, P94, P95, P98 Experiment (n=21) P2, P4, P7, P12, P17, P19, P21, P23, P27, P29, P38, P39, P54, P55, P63, P64, P65, P71, P80, P84, P90 Observation (n=10) P7, P8, P11, P20, P30, P49, P50, P80, P81, P93 Causal Mapping(n=1) P9 Documentation(n=4) P7, P30, P42, P50 Survey(n=8) P11, P34, P43, P57, P70, P72, P86, P91 Interview(n=20) P11, P15, P22, P25, P26, P28, P30, P32, P41, P49,P50, P54, P56, P72, P76, P77, P79, P82, P88, P93 Sample Analysis(n=2) P16, P87 ML subset selection (n=1)P18 Lit Analysis(n=22) P5, P6, P10, P14, P22, P24, P25, P28, P36, P37, P44, P45, P60, P61, P67, P68, P69, P73, P75, P78, P83, P96, P97 Workshop(n=3) P22, P28, P47, P58, Focus Groups(n=4) P11, P22, P31, P47 Questionnaire(n=2) P32, P35 Prediction Markets (n=1)P39C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 9primary studies, with 69 studies categorised under it. There was less range between the other types, expert systems having 11, machine vision withve and NLP at 6. Robotics was the least common with 3 studies having it as the primaryocus. The studies corresponding to each category can be seen on Table 12. There were aew studies that couldn't be said to haveocused on any one category specically, instead looking at AI in the broader sense. These 14 studies that could not be sorted cleanly into theramework were simply categorised as“other”. 4. 2. 6. RQ3: What is the business value oAI? As noted by Davenport and Ronanki (2018), within IS, it may be more useul to look at AI through the lens oits business capabilities rather than its technologies. To that end. AI can be narrowed down to Table 10 Contribution type (adaptedrom Shaw (2003),Paternoster et al. (2014) ). Title Description Framework/Method/ Technique The contribution othe study is a particularramework, method, or technique used toacilitate the construction and management osotware and systems. Guidelines A list oadvice or recommendations based on synthesis o the obtained research results. Lessons Learned The set ooutcomes directly based on the research results obtainedrom the data analysis. Model The representation oan observed reality in concepts or related concepts ater a conceptualisation process. Tool A technology, program, or application that is developed in order to support dierent aspects oinormation Advice/Implication A discursive and generic recommendation based on opinion. Table 11 Contributions across studies. Contribution Primary papers Framework/ Method P1, P13, P18, P19, P20, P21, P22, P24, P28, P29, P37, P43, P46, P48, P53, P55, P63, P64, P73, P74, P75, P78, P85, P92, P94, Guidelines P4, P5, P8, P16, P35, P59 Lessons Learned P3, P11, P12, P14, P15, P17, P25, P26, P30, P31, P32, P33, P34, P36, P39, P42, P44, P45, P49, P50, P56, P57, P58, P66, P67, P68, P70, P71, P72, P77, P80, P82, P84, P86, P89, P90, P91, P93, P98 Models P23, P38, P40, P41, P62, P79, P97 Tool P2, P9, P27, P65, P76, P81, Advice/ Implication P6, P7, P10, P47, P51, P52, P54, P60, P61, P69, P83, P87, P88, P95, P96 Fig. 5. Contribution types oprimary studies. Fig. 6. Types oAI. Table 12 Primary publications mapped to AI type. AI type Primary studies Machine Learning(n=68) P1, P3, P4, P5, P9, P13, P14, P15, P17, P18, P19, P21, P22, P23, P27, P29, P33, P34, P36, P37, P38, P39, P40, P42, P45, P46, P47, P48, P49, P50, P51, P52, P53, P55, P56, P57, P62, P63, P64, P65, P67, P73, P74, P77, P78, P79, P80, P82, P83, P86, P87, P88, P90, P91, P92, P93, P94, P95 Machine Vision(n=5) P2, P20, P25, P30, P43 Natural Learning Process (NLP) (n=6)P7, P28, P32, P54, P84, P89 Expert Systems(n=11) P8, P12, P16, P24, P31, P35, P44, P81, P85, P96, P97 Robotics(n=3) P11, P66, P72 Other(n=14) P6, P10, P26, P41, P58, P59, P60, P61, P68, P69, P70, P71, P75, P76C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 10support three business needs: (i) Process Automation, automating business processes, (ii) Cognitive Insight, gaining insight through data analysis, and (iii) Cognitive Engagement, engaging with customers and employees ( Davenport&Ronanki, 2018 ). In this study, we map the 98 primary studies to these three business needs (see Table 13 ). We acknowledge that it is possible thator some othese studies using AI could result in more than a single value type; however, to avoid complexity, they were mapped to the most relevant category. Fig. 7. shows that the most common value typeor AI was process automation (47 studies),ollowed by cognitive insight (32 studies). Cognitive engagement was reported with the lowest number ostudies (17 studies). 5. Discussion This section summarises thendings othe SLR and highlights some areas that research to date hasocused and the keyndingsrom these studies. It is thenollowed by a discussion on the theoretical contribu-tions and implicationsor practice. The overall goal is to uncover themes that are relevantor research and practice and identiy areas which warranturther research. This section will discuss relevant insights we oundrom the literature, starting with the lack ocohesion around the denition oAI, the resurgence oAI interest and research in recent years, the specic contribution types oAI literature, and the dispro-portionateocus on machine learning and process automation. In this study we conducted a SLR that provides a comprehensive overview on AI in IS related studies. By using a systematic literature review, we identied, classied, and analysed 1877 studies on AI and ML in IS that were published between 2005 and 2020. Othese, 98 were identied as primary studies, ater a rigorousltering process. To un-derstand theundamentals oAI in IS we examined and studied the ar-ticles based on studies by year, publication channel, research methods used, and their contribution to IS contributions research. Prior to commencing this task however, we had to consider the problem that the denitions oarticial intelligence were largely varied and ambiguous. 5. 1. Lack ocohesive defnition oAI This study identied a lack ocohesion when dening AI, with as many as 28 denitions being used is the respective studies. While the background research elaborated in section 2 shows this is not uncom-mon when concerning AI, it raises a concern that there is a high risk that IS studies on AI could experience a lack ocumulative building o knowledge ( Fitzgerald&Adam, 2000 ). This resonates with the issue o 'ragmented adhocracy', which has previously overshadowed IS research ( Banville&Landry, 1989; Hirschheim, Klein,&Lyytinen,1996 ). The most common denitionor AI seen was derivedrom Russel& Norvig, though the specic edition varied,ollowed by Le Cun et al. (2015) and Rai et al. (2019) with two occurrences each. Looking at the denitions gatheredrom the primary studies, there seems to be a trend where AI is dened more in what itscapabilitiesare rather than strictly dening what itis. Russel and Norvig (2020) dened it as something that “enables the machine to exhibit human intelligence, including the ability to perceive, reason, learn, and interact, etc. ” or example, while Stone et al. (2016) reers to it as“a science and a set ocomputational technologies that are inspired by—but typically operate quite dierentlyrom—the ways people [... ] sense, learn, reason, and take action”. This means that the denitions oAI oten end up being quite similar, even ithey are taken rom a separate source. While dening AI is outside the scope othis study, the most robust denition oAI in the context oIS research is provided by Rai et al. (2019) who dene it as“the ability oa machine to perorm cognitiveunctions that we associate with human minds, such as perceiving, reasoning, learning, interacting with the environment, problem solving, decision-making, and even demonstrating creativity”. However, there may be issues with how IS research is approaching the dening oAI as there seems to be a believe that there is a true (i. e., real, natural) meaning o“intelligence” that AI research projects should abide by, at least among those that consider dening AI. Yet“intelli-gence” by its dictionary denition today wasormed long beore AI, and thereore strictly about human intelligence, where the various levels (conscious and subconscious) are unied, which is not the caseor computer intelligence. For human intelligence, structure, behaviour, capability, andunction are all relatively unied, whileor AI these aspects are commonly pursued in dierent ways towards dierent goals. For example, an AI created by mimicking thestructureoa human brain and one created byocusing on imitating humanbehaviourwould be end up being very dierent both in its method and its result. Additionally, human intelligence is developed under certain evolutionary and bio-logical restrictions, which are essentialor human, but not reallyor intelligence in general. Ater all,“Articial Intelligence” should not be taken to mean“Articial Human Intelligence”, since“Intelligence” should be a more general notion than“Human Intelligence”. Another major concern is the number ostudies concerning or using someorm oAI that do not dene it at all. Othe 98 primary studies, 54 didn't dene AI at all. These are primarily the studies that use AI as part othe methodology otheir study, oten in the case ousing some type o machine learning in pursuit othat studies objective. While it is not easible to ask every researcher, who uses a machine learning technique in their research to dedicate an entire section to AI, the lack osucient detail seen in many othe primary studies showcases a possible lack in the cumulative building oknowledge. A related concern is the 7 studies who dened AI by themselves, without sourcing any part it. This is something to be discouraged as much as possible, as it is important both or the individual researchers andor theeld as whole to have knowledge built upon clearly and accurately. Table 13 Reported value types oAI studies. Value type Description ( Davenport&Ronaki, 2018 )Primary studies Process automation (n=49)Automating business processes P1, P5, P6, P9, P10, P11, P12, P13, P14, P15, P16, P17, P18, P25, P26, P28, P29, P30, P31, P33, P34, P36, P39, P40, P41, P43, P45, P47, P50, P52, P55, P56, P60, P61, P62, P64, P68, P69, P70, P76, P77, P78, P82, P83, P86, P88, P90, P93, P97 Cognitive insight (n=32)Gaining insight through data analysis P3, P4, P8, P19, P20, P21, P22, P23, P24, P27, P37, P42, P48, P51, P53, P57, P58, P59, P63, P65, P73, P74, P75, P79, P80, P81, P87, P91, P92, P95, P96, P98 Cognitive engagement (n=17)Engaging with customers and employees P2, P7, P32, P35, P38, P44, P46, P49, P54, P66, P67, P71, P72, P84, P85, P89, P94Fig. 7. Number oprimary studies by value type. C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 115. 2. The resurgence oAI in recent years Ourndings show that while there was a lack ostudies in therst decade othe relevant time, there has been a resurgence in recent years. Over halothe primary studies were published in 2019 and 2020. Our ndings also show that ICIS published the highest number ostudies. ICIS published 41 studies to AI in Inormation Systems in comparison to the 24 studies published in ECIS. IJIM published 14 primary studies, making them the most published regarding AI in IS journals used in the systematic review. As previously shown in Fig. 4., there was a resurgence in interest in AI in 2017, beore it seemed to really expand in 2019 and 2020. It's already been noted that AI research tends to swing between its“winters” and“summers”, and it seems clear that we have headed into one oits summer periods in recent years. In 2016, the Economist declared in its June Special Report ( The Economist, 2016 ) that“Ater manyalse starts, articial intelligence has taken o. ” This wasollowed by reports released by high level policy makers such as the US and UK governments onuture directions and strategyor AI, noting its immense potential but also questions on accountability and use ( Government Oceor Sci-ence, 2016; U. S. National Science and Technology Council, 2016 ). This surge ointerest in the use oAI among businesses and policy makers naturally resulted in a matching surge in interest among re-searchers, who now have much more to study due to AIs ever increasing use. This resurgence in AI can be put down to aew reasons. One othe biggest is a convergence oadvances in machine learning and big data and graphics processing units (GPUs) which made supervised learning rom large datasets much more practical ( Morris, Schleno,&Sriniva-san, 2017 ). This surge in interest in AI seems to correlate with our own ndings, as researchers now have much more access to empirical evi-dence and AI in use with the advent othis“summer”. In the IS com-munity specically, the increasedeasibility obuilding empirical modelsrom experimental data and using these models to make pre-dictions (the goal opredictive analytics using big data) is likely another reasonor the resurgence ointerest. However, despite the abundance o research related to machine learning, most owhat weound in the literature was related to its technological use in specic domains, whether that domain be healthcare, manuacturing etc. There is a relative lack oresearch into societal or governmental implications o machine learning and its recent advancements. 5. 3. Contribution mismatch For identiying the contribution types that AI studies have made in IS in the past 15 years, the primary studies were categorised using a ramework adaptedrom Shaw (2003) and Paternoster et al. (2014). The most prominent contribution type among the 98 studies was categorised as“lessons learned” with 40 studies, closelyollowed by“methods” at 26 and to a lesser extent“advice or implication” at 15. While thesendings do show a steady gathering ocumulative knowledge in relating to AI in IS, it also showcases a lack ostudies in relation to (i) tools and (ii) models. This is somewhat surprising, as this means there seems to be little interest in IS in researching new and innovative tools relating to AI. Instead, the bulk othe research seems to beocused on using types oAI, primariliy machine learning, to create a particularramework, method, or technique and use that toacilitate the construction and management osotware and systems. It seems likely that the reasonor this lack o ocus on tools or methods is related in some ways to the reasonsor the AI resurgence itselnoted in section 5. 2. That is, the advancement otechnology making supervised learningrom datasets and machine learning much moreeasible means that researchers are nowocused on this neworay, and“tools” and“methods” are let to languish because it seems the resurgence doesn't lit up all aspects oAI equally. 5. 4. Focus on machine learning In terms ostudying the categories oAI used,ollowing therame-work provided by Dejoux and L ´eon (2018), machine learning wasound to be overwhelmingly prominent in the primary studies, with 69 othe 98alling under it. Expert systems were the second most studies type o AI with 11 studies categorised as using it. Robotics saw the least use among the primary studies, with 3 studies concerned with it as its main ocus. This machine learning dominance seems to be due to, at least in part, the wide variety ocontexts in which machine learning is useul in comparison to the other types. For example, many othe primary studies dierentiated themselves by using a“machine learning approach” (Meyer et al., 2014 ;Chatterjee, Saeedar, Toangchi,&Kolbe, 2018 ). In addition, therequent appearance omachine learning as the AI appli-cation used in this paper can be attributed to theact that it concerns a very broad spectrum opotential applications. Thereore, while the prominence omachine learning may be understandable, it is still a concern. It means there is a relative dearth in studies concerning the other categories under AI, especially robotics. Google Trends show that popularity othe search termor machine learning has surpassed the popularity oAI by almost twice as much ( Google Trends, 2020 ). Additionally, the advances in hardware such as GPUs that has made AI much moreeasible in recent years seems to disproportionallyavour machine learning, which means moreocusrom industry and research. For example, text mining relatively inexpensive and can result in a wealth oinormation idone correctly ( He, Zha,&Li, 2013 ) while there has been some research done on the critical successactors odata-mining that is lacking in other areas ( Bole, Popovic, Zabkar, Papa,& Jaklic, 2015 ). This means that some othe other types oAI have received less researchocus, though all othem can oer much both practically and academically. For researchers, this seems to have resulted in many more studies approaching research into their own domains using what is oten simply called“a machine learning approach” with P1 ( Meyers et al., 2014 ), P13 ( Yin, Langenheldt, Harlev, Mukkamala,&Vatrapu, 2019 ) and P18 ( Buettner et al., 2019 ) just being aew othe many examples. From our ownndings here, the papersocused on natural language processing (NLP) have chatbots and similar virtual assistants as a major ocus. However, they seemocused onrontacing interactions between these agents and customers, and less on actual employees working be-side these agents. As chatbots and virtual agents rapidly advance in complexity, there needs to be more research into not just the eects on customers interacting with them, but the human-agent interaction othe people workingbesideandwiththem as well. The studies weound that used expert systems (ES) seemed toavour using a hybrid knowledge base using a variety oAI systems, rather than the“classical” method ousing just one or more human experts ( Kunz, Stelzner,&Williams, 1989 ). However, it has been noted by other studies that this shit hasn't resulted in a greater impact than earlier systems, despite the advances in technology. This was noted by ( Wagner, 2017 ), who also theorised that the reasonor this lack oimpact is that the earlier developers were able to capitalise on the 'low hangingruit' that had bigger impactor organisations. 5. 5. Focus on process automation Regarding the business value oAI in the primary studies, the results aren't immensely surprising, and inact align with thendings o Davenport and Ronanki (2018). Just as with ( Davenport&Ronanki, 2018 ), process automation was the most reported business valueound because oAI use. This is likely due to process automation being both the least expensive and the easiest to implement othe three value types discussed here ( Davenport&Ronanki, 2018 ). Adopters oRPA have noted the automation can radically transorm back oces, delivering much lower costs while improving service quality, and decreasing de-livery times, as well asreeing up employeesrom tedious tasks so they C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 12canocus on more important, challenging, and varied work ( Lacity, Willcocks,&Andrew, 2015 ). For the primary studies, this oten took the orm othe authors using someorm oautomated ML data mining. Another example would be P36, a paper which studied the application o machine learning in decision support systems andound that the pri-mary value was in acquiring and rening the knowledge used as the basis othe decision, rather than it make decisions themselves, noted in the studyor being much more dicult ( Merkert, Mueller,&Hubl, 2015 ). Cognitive insight was the second most common type oderived business value among the primary studies and was described by Davenport and Ronanki (2018 ) as“analytics on steroids”. Organisations that used such applications oAInd their value in perorming and enhancing tasks only machines can do. Among the primary studies o this study, P55 would be a good standout othis, studying the use o deep learning to enhance customer targeting eectiveness ( Zhang& Luo, 2019 ). The paper asserted that given some pilot tests deep learning models would be superior in sales perormance compared to the more common industry practices otargeting customers by past purchase requency or spending amount. These activities involve data analysis at such speeds and at volumes that no human would be possible to process. Cognitive engagement applications oAI were the value attributed to the least number oprimary studies within our sample. This is likely partly due to the cognitive engagement dealing more with customers, and businesses being more conservative with customeracing technol-ogy ( Davenport&Ronanki, 2018 ). The idea ohavingrustrating con-versions with a chat-bot that just can't seem to understand what the user is saying still quite strong in the public consciousness, never mind the high-proleailures such as Taybot ( Badjatiya, Gupta, Gupta,&Varma, 2017 ). Ourndings provide an overview othe current state oresearch when it comes to AI use in the organisational setting, but also helps to draw some interesting pointsoruture research. One othe main ndings in our synthesis concerned the lack odenitions when it comes to AI in empirical studies. Since AI applications cover a breadth o dierent techniques, technologies, and set a dierent set ore-quirements on data, inrastructure and leveraging them in the organ-isational setting it is important thatuture research accuratelyrames the denition oAI that is used. Apartrom enabling a better comparison between empirical works, clearly articulating denitions can also allow or a better understanding othe assumptions and constraints that characterise the body owork. Adding to the above, the dierent applications oAI largely dictate the type obusiness value that can be expected. This is a point that is mentioned also in the article o Davenport and Ronanki (2018) but is largely overlook in empirical studies. Providing exact denitions on what type oAI application is studied is criticalor business value research. In addition, it is important that studies dene the exact use and application othe technology since this has an important bearing on how business value is realised. This point also relates heavily to the choice o theory used to support business value generation, as well as the context in which these technologies are deployed. Specically, on the use otheories in studying business value, much othe work that has been published to date remains untheoretical. This poses a major issue as the boundary conditions and context in which AI applications are studied are not grounded on established theoretical rameworks. It also makes the comparison betweenndings and the identication ocomplementarities much more dicult. Adding to the above, major themes that have been under-reached remain dicult to identiy with an absence otheoretically grounded work. For instance, there is limited work on the diusion and assimilation oAI application in organisationsollowing longitudinal studies. This creates a large gap in our understanding ohow AI applications are gradually assimilated in operations and how business value may evolve depending on the dierent stages omaturity. 5. 6. Implicationsor practice Ourndings, apartrom their research relevance also raise some important practical implications. Specically, our analysis documents the types oAI applications that are most pursued by organisations and thereore ohighest interest to researchers. The specic applications and technologies oAI that are mostly researched provide practitioners some indication about theuture deployments and common technologies in organisations. Theact that machine learning applications are the most researched technology within the AI domain provides some indication about whereuture investments should be directed, as well as the ex-pected type obusiness value that they can deliver. Having such insight can allow IT managers to start experimenting with such techniques within their organisations and making appropriate investments to gradually deploy such solutions in business areas where they could be o high value. In addition, the review ostudies points out to those that can enable practitioners to obtain important lessons learnedrom deployments o AI technologies, identiy ways in which methods have been applied and what common challenges emerge, as well as identiy those that present general advice and best practices. The broad and extensive literature on AI in organisational settings make it challengingor many practitioners to identiy empirical studies that are ovalue to them. With the synthesis o ndings and the presentation ostudies based on a thematic catego-risation, practitioners are more easily able to identiy those studies that contribute to the challenges they and their organisationsace when it comes to AI deployments. In section 2. 2, we noted other SLRs conducted in IS concerning articial intelligence. Rzpeka and Berger (2018) consolidated research streams within IS research that had previously been treated separated and aggregated insights regarding the interaction with dierent AI-enabled system types. Homann et al. (2019) ound that nearly every step in the radiology value chain could be improved with the use o machine learning. Borges et al. (2021) ound that the strategic use oAI had not been well explored yet and created a preliminary conceptual ramework to aid managers in exploring that. Karger (2020) was arst attempt to investigate how block chain and AI could combine. This study took a step backrom any specic industry domain to research how AI and machine learning was being dened and used in a broader level, uses aramework adaptedrom Shaw (2003) and Paternoster et al. (2014) to see the contributions to literaturerom AI studies in IS and creates a research agendaoruture research. 6. Future research agenda In theollowing section, we critically evaluate the literature related to our research questions and highlight potential gapsorurther study to identiy the opportunitiesoruture AI research and thusull the ourth andnal aim othis paper. We develop an agenda o uture research that buildrom the identied gaps. This research agenda is presented in Table 14 and was primarilyormatted to correspond to the ramework provided by Dejoux and L ´eon (2018) that has been used throughout this study, with the addition otwo key areas that warrant urther research. AI is seeing a small consolidation in how it is dened; as something that exhibits human intelligence, which can be seen in the denitions used by the papers in section 4. 1. But the subject and denition o human intelligence is something that is still heavily debated even now. It has been noted that AI technology has tended to“become a somewhat broad church where manyorms oautomation and limited intelligent ma-chines are labelled as AI” ( Dwivedi et al., 2021 ). There is a gap thereor researchers to give more clarity in dening AI, even ithat means redening it awayrom traditional human intelligence. Section 4. 2. 1 also points to a resurgence in interest in AI in recent years, though many othe studies here are heavilyocused on the technology and peror-mance aspect oAI. More research should be done on the societal and C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 13personal eects these recent advances will have on people, both in the workplace and their everyday lives. Researchers are increasingly using machine learning as a major component otheir methodology when researching various topics, but currently there is a gap where much othis research couldurther expand on how and why they are using machine learning the way they are. For expert systems, there seems to be a shit awayrom the "tradi-tional" style oES with just aew experts managing a system to a more hybrid knowledge base that uses a variety oAI systems ( Wagner, 2017 ). This increased complexity may also mean it will take more time and eort than usual to see a return on the time and resources invested. Research into theull aspects ohow service robots could potentially aect business and people is still lacking. Wider debate is needed into the interactive and psychological elements orobot-human interaction, especially in the long term, with some studies specically showing that the strategic use oAI technologiesor customer and employee engagement has not been well exploited yet ( Borges et al., 2021 ;Gursoy, Chi, Lu,&Nunkoo, 2019 ). A specic domain that is currently lacking in researchor robotics is technophobia; previous research has examined theear ocomputers and have not accountedor new and evolvingtechnologies such as robots ( Sinha, Singh, Gupta,&Singh, 2020 ). Research into extensive interactions between advanced chatbots and humans is still immature, so researchers should take advantage othe increase in NLP capability and the usage oits technologies to dourther research. Finally, machine vision doesn't seem to be taking advantage o some othe recent technological advances seen in the new studies in machine learning, so there's a gap thereor researchers to see iand how machine learning could be improved. Additionally, much oIS literature recognises that IS alone is ineective in generating value, so comple-mentary assets are key to realising valuerom IS ( Shea, Dow,&Chong, 2019 ). AI is one othese complementary assets with potentialor transormative value in IS ( Nishant, Kennedy,&Corbertt, 2020 ). There is potentialor AI in all itsorms. Despite the interest in AI in recent years, there remains gaps in knowledge. However, AI does tend to encompass a broad array oideas and practises, rangingrom the spe-cics othe technology used to even how it is dened. Manyorms o automation, machine learning and intelligent agents are thus labelled as AI. However, despite the great strides in AI noted in the, so called “strong” AI doesn't seem like it will be made a reality within theore-seeableuture ( Kurzweil, 2005 ). Theuture agenda oAI seems set on theurther advancement o“weak” AI where specic tasks and decisions are attributed to machines, especially as much othe current research is ocused on industry specic uses oAI i. e., AI in healthcare, AI in manuacturing etc. This leaves a gapor researchers to consider the use oAI in other domains, such as agriculture sustainability ( Nishant, Kennedy,&Corbertt, 2020 ) and the public sector ( Dennehy, 2020 ; Abubakar, Behravesh, Rezapouraghda,&Yildiz, 2019 ). From thendings and analysis othis SLR, it highlights a needor a more coherent working denition oAI among academia. To improve the coherence and eciency oresearch and communication, it is better to make our working denitions explicit. 7. Conclusion This systematic literature review study provides a structured un-derstanding othe state-o-the-art oAI research in IS. This was achieved by identiying 98 primary studies out o1877 related AI articles over a teen-year period (2005-2020) and analysed them with respect to (i) denitions oAI, (ii)requency opublication by year, (iii) publication channels, (iv) research method and data collection type, (v) contribution type, (vi) type oAI and (vii) business value. A clearnding emergingrom this systematic literature review is the need to (i) increase the number origorous academic studies on AI, especially regarding tools and models, (ii) be more detailed on the denition oAI used in studies, even when it is not theocus, and (iii) build on cumulative knowledge. Research on AI in IS is still largely unexplored. While there is a relatively sizable amount oliterature concerning AI in some way, a comprehensive review owhat is known about AI in IS is lacking. This is especially trueor the way AI is dened in IS, which is still disparate. This study examines the body oknowledge about AI in IS. This work has developed one othe veryew SLRs on AI in IS and has provided a structured analysis otrends and gaps in theeld. The study provides new insights to theeld oIS through the utilisation oconceptions oAI denition, mapping activities to AI, and value relating to AI. We identied gaps in knowledge in the context oAI research and IS, which provides a starting pointor IS researchers and IS practitioners to advance the socio-technical knowledge surrounding AI. Thus, we make a calloruture IS studies to examine AI, specically to how AI is dened in contemporary IS research. Acknowledgements This research wasunded through a scholarship awarded by the Business Inormation Systems discipline, J. E. Cairnes School oBusiness &Economics, NUI Galway, Galway, Ireland. Table 14 Future research agendaor AI. Title Research agenda description Future Research Questions AI denition Lack oconsensus around the denition oAIIs comparing AI to human intelligence the most eective way oadvancing AI research? How can arst principles approach be used to dene a more contemporary denition oAI? Resurgence o interest Overocus on the technology and perormance aspects oAIWhat are the societal and personal impacts othe recent advances in AI? What can researchers and regulators do to keep up with the speed othese advances? Machine learning Increase in use omachine learning as a methodology among researchers. How can a researcher measure the eectiveness o their machine learning approach? Expert systems (ES)Moverom“classical” ES to a more hybrid knowledge base. Is the change to a more hybrid knowledge base o expert systems more eective than the “classical” style? Iso, does the value added by this new style outweigh the resources and time spent on adapting to it? Robotics Eects oextended use o advanced service robots on people is still relatively undeveloped. What are the impacts othe use oservice robots on people, both those they are servicing and the people working alongside them? What are the long-term psychological impacts on the increased use oservice robots, both on an individ-ual and societal level? Natural Language Processing (NLP)Chatbots and intelligent agents have made great advancements in recent years, while the eects o these advancements still need to be studied. How can we quantiy the value omore advanced chatbots and intelligent agents? Machine vision Machine vision seems to be lagging in advances compared to strides made in other AIunctions. How can the recent advances in AI and hardwareurther improve the use omachine vision?C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 14Reerences Abubakar, A. M., Behravesh, E., Rezapouraghda, H., & Yildiz, S. B. (2019). Applying articial intelligence technique to predict knowledge hiding behavior. International Journal oInormation Management, 49, 45-57. Agarwal, S., Kumar, S., & Goel, U. (2019). Stock market response to inormation diusion through internet sources: A literature review. International Journal oInormation Management, 45, 118-131. Ali, O., Shrestha, A., Soar, J., & Wamba, S. F. (2018). Cloud computing-enabled healthcare opportunities, issues, and applications: A systematic review. International Journal oInormation Management, 43, 146-158. Allen, J. F. (1998). AI growing up: The changes and opportunities. AI Magazine, 19(4), 13-23. Austin, P., Tu, J., Ho, J., Levy, D., & Lee, D. (2013). Using methodsrom the data-mining and machine-learning literatureor disease classication oheartailure subtypes. J. Clin. Epidemiol., 398-407. Badjatiya, P., Gupta, S., Gupta, M.,&Varma, V. (2017). Deep learningor hate speech detection in tweets. 26th International World Wide Web Conerence. Ballew, B. (2009). Elsevier's Scopus®database. Journal oElectronic Resources in Medical Libraries, 6(3), 245-252. Banville, C., & Landry, M. (1989). Can theeld oMIS be disciplined?Communications o the ACM, 32(1), 48-60. Berente, N., Gu, B., Recker, J.,&Santhanam, R. (2019). Managing AI. MIS Quarterly. Bhatnagar, S., Alexandrova, A., Avin, S., Cave, S., Cheke, L., Crosby, M., & Hernandez-Orallo, J. (2018). Mapping intelligence: Requirements and possibilities. In V. C. Müller (Ed. ),Philosophy and theory oartifcial intelligence 2017(pp. 117-135). Berlin: Springer. Bole, U., Popovic, A., Zabkar, J., Papa, G., & Jaklic, J. (2015). A case analysis o embryonic data mining success. International Journal oInormation Management, 35 (2), 253-259. Borges, A. F., Laurindo, F. J., Spínola, M. M., Gonçalves, R. F., & Mattos, C. A. (2021). The strategic use oarticial intelligence in the digital era: Systematic literature review anduture research directions. International Journal oInormation Management, 57, Article 102225. Brachman, R. J. (2006). AA)AI—more than the sum oits parts, 2005 AAAI Presidential Address. In AI Magazine, 27pp. 19-34). Brynjolsson, E., & Mc Aee, A. (2014). The second machine age: Work, progress, and prosperity in a time obrilliant technologies. New York: WW Norton&Company. Buettner, R., Frick, J.,&Rieg, T. (2019). High perormance detection oepilepsy in seizure-ree EEG recordings: A novel machine-learning approach using very specic epilectic EEG sub-bands. ICIS 2019 Proceedings. Busalim, A. H., & Hussin, A. R. (2016). Understanding social commerce: A systematic literature review and directionsorurther research. International Journal o Inormation Management, 36(6), 1075-1088. Bush, V. (1945, July 1). As we may think. The Atlantic Magazine. Carbonell, J., Michalski, R., & Mitchell, T. (1983). An overview omachine learning. In R. Michalski, J. Carbonell, & T. Mitchell (Eds. ),Machine Learning: An Artifcial Intelligence Approach. Palo Alto, Caliornia: TIOGA Publishing Co. Cepolina, F.,&Muscolo, G. (2014). Design oa robotor hygienization owalls in hospital environments. Proceedings othe 41st International Symposium on Robotics. Munich. Chatterjee, S., Saeedar, P., Toangchi, S.,&Kolbe, L. (2018). Intelligent road maintenance: a machine learning approachor surace deect detection. ECIS 2018 Proceedings. Chi, M., Huang, R., & George, J. F. (2020). Collaboration in demand-driven supply chain: Based on a perspective ogovernance and IT-business strategic alignment. International Journal oInormation Management, 52, Article 102062. Collins, J., Youngdahl, B., Jamison, S., Mobasher, B.,&Gini, M. (1998). A market architectureor multi-agent contracting. Minneapolis: K. Sycara and M. Wooldridge (eds). Coombs, C. (2020). Will COVID-19 be the tipping pointor the Intelligent Automation o work? A review othe debate and implicationsor research. International Journal o Inormation Management, 55, Article 102182. Cross, S. (2003). Agency, contract and intelligent sotware agents. Int. Rev. Law, Comput. Technol., 175-189. Davenport, T. H.,&Ronanki, R. (2018, January). Articial Intelligenceor the Real World. Harvard Business Review. De Canio, S. J. (2016). Robots and humans-complements or substitutes?J. Macroecon., 280-291. Dejoux, C., & L ´eon, E. (2018). M ´etamorphose des managers(1st ed. ). France: Pearson. Dennehy, D. (2020). Ireland post-pandemic: Utilizing AI to kick-start economic recovery. Cutter Business Technology Journal, 33(11), 22-27. Deng, L., & Yu, D. (2014). Deep learning: methods and applications. Found. Trends Signal Process., 197-387. Dingsøyr, T., & Dybå, T. (2008). Empirical studies oagile sotware development: A systematic review. Inormation and Sotware Technology, 50, 833-859. Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Articial intelligenceor decision making in the era oBig Data-evolution, challenges and research agenda. International Journal oInormation Management, 48, 63-71. Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T.,... Medaglia, R. (2021). Articial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agendaor research, practice and policy. International Journal oInormation Management, 57, Article 101994. Fink, A. (2005). Conducting research literature reviews: From the internet to paper(5th ed. ). Los Angeles: Sage. Fitzgerald, B., & Adam, F. (2000). The status othe inormation systemseld: Historical perspective and practical orientation. Inormation Research, 5(4). Goodellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. Google Trends. (2020). Google Trends. Retrieved 10 11, 2020,rom 〈https://trends. google. com/trends/explore?date=all&q=machine%20learning,artiicial%20intell igence〉. Government Oceor Science. (2016). Articial intelligence: opportunities and implicationsor theuture odecision making. London. Günther, W., Mehrizi, M., Feldberg, F., & Huysman, M. (2017). Debating big data: A literature review on realizing valuerom big data. J. Strateg. In. Syst., 191-209. Gupta, S., Kar, A. K., Baabdullah, A., & Al-Khowaiter, W. A. (2018). Big data with cognitive computing: A reviewor theuture. International Journal oInormation Management, 42, 78-89. Gursoy, D., Chi, O. H., Lu, L., & Nunkoo, R. (2019). Customer acceptance oarticially intelligent (AI) device use in service delivery. International Journal oInormation Management, 49, 157-169. He, W., Zha, S., & Li, L. (2013). Social media competitive analysis and text mining: A case study in the pizza industry. International Journal oInormation Management, 33(3), 464-472. Hearst, M. A., & Hirsh, H. (2000). AI's greatest trends and controversies. IEEE Intelligent, 8-17. Hengstler, M., Enkel, E., & Duelli, S. (2016). Applied articial intelligence and trust—the case oautonomous vehicles and medical assistance devices. Technol. Forecast. Soc. Change, 105-120. Hern´andez-Orallo, J. (2017). The measure oall minds: Evaluating natural and artifcial. Cambridge: Cambridge University Press. Hirschheim, R., K. Klein, H., & Lyytinen, K. (1996). Exploring the intellectual structures oinormation systems development: A social action theoretic analysis. Accounting, Management and Inormation Technologies, 6(1-2), 1-64. Homann, P., Oesterle, S., Rust, P.,&Urbach, N. (2019). Machine learning approaches along the radiology value chain-Rethinking value propositions. Stockholm: ECIS 2019 Proceedings. International Federation oRobotics. (2016). Service Robots. Retrievedrom IFR International Federation oRobotics: https://ir. org/service-robots/. Jain, A., Ross, A., & Prabhakar, S. (2004). An introduction to biometric recognition. IEEE Transactions on Circuits and Systemsor Video Technology, 14(1), 4-20. Jarrahi, M. (2018). Articial Intelligence and theuture owork: Human-AI symbiosis. Business Horizons, 61(4), 577-586. Jarrahi, M. (2018). Articial intelligence and theuture owork: Human-AI symbiosis in organizational decision making. Business Horizons, 64(4), 577-586. Jennings, N. R., Sycara, K., & Wooldridge, M. (1998). Roadmap oagent research and development. Auton. Agent. Multi Agent Syst., 7-38. Juniper Research. (2018). AI in retail. segment analysis, vendor positioning&market. Retrieved February 2021,rom 〈https://www. juniperresearch. com/researchstore /ntech-payments/ai-in-retail-research-report〉. Karger, E. (2020). Combining Blockchain and Articial Intelligence-Literature Review and State othe Art. ICIS 2020 Proceedings. Khashman, A. (2009). Application oan emotional neural network toacial recognition. Neural Computing and Applications, 18(4), 309-320. Kitchenham, B. (2004). Proceduresor perorming systematic reviews(pp. 1-26). Keele, UK: Keele Univ. Kitchenham, B.,&Charters, S. (2007). Guidelinesor perorming systematic literature reviews in sotware engineering. In Technical report, Ver. 2. 3 EBSE Technical Report. EBSE. Kitchenham, B., Budgen, D., & Brereton, O. (2011). Using mapping studies as the basis orurther research-A participant-observer case study. Inormation and Sotware Technology, 53(6), 638-651. Kolbjørnsrud, V., Amico, R., & Thomas, R. J. (2017). Partnering with AI: how organizations can win over skeptical managers. Strategy Leadersh., 37-43. Kunz, J. C., Stelzner, M. J., & Williams, M. D. (1989). From classic expert systems to models: Introduction to a methodologyor building model-based systems. Studies in Computer Science and Artifcial Intelligence, 5, 87-110. Kurzweil, R. (2005). The singularity is near. New York: Viking. Lacity, M., Willcocks, L., & Andrew, C. (2015). Robotic process automation: Mature capabilities in the energy sector. LSE Research Online Documents on Economics. Langley, P. (2011). Articial Intelligence. AISB Quarterly. Le Cun, Y., Bengio, Y., & Hinton, G. (2015). Deep Learning. Nature, 521, 436-444. Leidner, D., & Kayworth, T. (2006). A review oculture in inormation systems research: toward a theory oinormation technology culture confict. MIS Quarterly, 30(2), 357-399. Levy, Y., & Ellis, T. (2006). A systems approach to conduct an eective literature review in support oinormation systems research. Inorming Science, 9, 181-212. Li, Y. (2017). Deep Reinorcement Learning: An Overview. ar Xiv preprint ar Xiv. Longoni, C., Bonezzi, A., & Morewedge, C. K. (2019). Resistance to medical articial intelligence. J. Consum. Res., 629-650. Lugano, G. (2017). Virtual assistants and sel-driving cars. Warsaw: 2017 ITST Proceedings. Majchrzak, A., Markus, M. L., & Wareham, J. (2016). Designingor digital transormation: lessonsor inormation systems researchrom the study oICT and societal challenges. MIS Quarterly, 40, 267-277. Manikandan, M., & Amsaveni, N. (2016). Management Inormation System research output: A scientometric study. International Journal oLibrary&Inormation Science, 5 (1), 21-27. Mc Carthy, J. (1958). Programs with common sense. Proceedings othe Symposium on Mechanisation, 77-84. C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 15Mc Carthy, J. (1988). Mathematical logic in articial intelligence. Dædalus, 117(1), 297-311. Mc Corduck, P. (2004). Machines who think: A personal inquiry into the history and prospects oartifcial intelligence(2nd ed. ). A. K. Peters. Mc Culloch, W. S., & Pitts, W. H. (1943). A logical calculus oideas immanent in neural activity. Bulletin oMathematical Biophysics, 5, 115-133. Merkert, J., Mueller, M.,&Hubl, M. (2015). A survery othe application omachine learning in decisiion support syystems. ECIS 2015 Proceedings. Mettler, T., Sprenger, M., & Winter, R. (2017). Service robots in hospitals: New perspectives on niche evolution and technology aordances. European Journal o Inormation Systems, 26(5), 451-468. Meyer, G., Adomavicius, G., Johnson, P., Elidrisi, M., W. A., R., Sperl-Hillen, J., & O'Connor, P. (2014). A machine learning approach to improving dynamic decision making. Inormation Systems Research, 25(2), 239-263. Mikale, P., Bourab, M., Lekakosb, G., & Krogstiea, J. (2019). Big data analytics andrm perormance: Findingsrom a mixed-method approach. Journal oBusiness Research, 98, 261-276. Min, H. (2010). Articial intelligence in supply chain management: theory and applications. Int. J. Logist. Res. Appl., 13-39. Minsky, M. (1958). The society omind. New York: Simon and Schuster. Monett, D., & Lewis, C. W. (2018). Getting clarity by dening Articial Intelligence-A Survey. In V. C. Müller (Ed. ),Philosophy and theory oartifcial intelligence 2017(pp. 212-214). Berlin: Springer. Morris, K., Schleno, C., & Srinivasan, V. (2017). A remarkable resurgence oarticial intelligence and its impact on automation and autonomy. IEEE Transactions on Automation Science and Engineering, 14(2), 407-409. Newell, A., & Simon, H. A. (1963). Gps, a program that simulates human thought. Computers and thought(pp. 279-293). Cambridge: MIT Press. Newell, A., & Simon, H. A. (1976). Computer science as empirical enquiry: Symbols and search. Communications othe ACM, 19, 113-126. Nichols, G. (2018). Robotics in business: Everything humans need to know. Retrieved rom ZDNet: https://www. zdnet. com/article/robotics-in-business-everything-huma ns-need-to-know/. Nilsson, N. J. (2009). The questor artifcial intelligence: A history oideas and achievements. Cambridge: Cambridge University Press. Nishant, R., Kennedy, M., & Corbertt, J. (2020). Articial intelligenceor sustainability: Challenges, opportunities, and a research agenda. International Journal oInormation Management, 53, Article 102104. Nissen, M. E., & Sengupta, K. (2006). Incorporating sotware agents into supply chains: Experimental investigation with a procurement task. MIS Quarterly, 30(1), 145-166. Okoli, C. (2015). A guide to conducting a standalone systematic literature review. Communications othe Associationor Inormation Systems, 37, 43-52. Paternoster, N., Giardino, C., Unterkalmsteiner, M., Gorschek, T., & Abrahamsson, P. (2014). Sotware development in startup companies: A systematic mapping study. Inormation and Sotware Technology, 56, 1200-1218. Petersen, K., & Wohlin, C. (2009). A comparison oissues and advantages in agile and incremental development between state othe art and an industrial case. Journal o Systems and Sotware, 82(9), 1479-1490. Petersen, K., Vakkalanka, S., & Kuzniarz, L. (2015). Guidelinesor conducting systematic mapping studies in sotware engineering: An update. Inormation and Sotware Technology, 64, 1-18. Petersen, K., Vakkalanka, S., & Kuzniarz, L. (2015). Guidelinesor conducting systematic mapping studies in sotware engineering: an update. Inormation Sotware Technology, 64, 1-18. Petticrew, M.,&Roberts, H. (2006). How to appraise the studies: An introduction to assessing study quality. In Systematic reviews in the social sciences: A practical guide (pp. 125-163). Plastino, E., & Purdy, M. (2018). Game changing valuerom articial intelligence: eight strategies. Strategy Leadersh., 16-22. PWC. (2019). Sizing the prize: exploiting the AI Revolution, What's the real value oAI or your business and how can you capitalise? Pw C's Global Articial Intelligence Study. Rai, A., Constantinides, P., & Sarker, S. (2019). Next-generation digital platorms: Towards human-AI hybrids. MIS Quarterly, 43, 3-8. Ransbotham, S., Fichman, R. G., Gopal, R., & Gupta, A. (2016). Special section introduction—Ubiquitous IT and digital vulnerabilities. Inormation Systems Research, 27(4), 834-847. Rekik, R., Kallel, I., Casillas, J., & Alimi, A. M. (2018). Assessing web sites quality: A systematic literature review by text and association rules mining. International Journal oInormation Management, 38(1), 201-206. Rodriguez-Aguilar, J. A., Martin, F. J., Noriega, P., Garcia, P.,&Sierra, C. (1998). Competitive Scenariosor Heterogeneous Trading Agents. Minneapolis: K. Sycara and M. Wooldridge. Rubik, B., & Jabs, H. (2018). Articial intelligence and the human bioeld: New opportunities and challenges. Cosmos and History, 14(1), 153-162. Russel, S., & Norvig, P. (1995). Artifcial Intelligence: A Modern Approach(1st ed. ). N. J. : Prentice Hall. Russel, S., & Norvig, P. (2010). Artifcial Intelligence: A Modern Approach(2nd ed. ). N. J. : Prentice Hall. Russel, S., & Norvig, P. (2016). Artifcial Intelligence: A Modern Approach(3rd ed. ). N. J. : Prentice Hall. Russel, S., & Norvig, P. (2020). Artifcial Intelligence: A Modern Approach(4th ed. ). N. J. : Prentice Hall. Rzpeka, C.,&Berger, B. (2018). User interaction with AI-enabled systems: A systematic review oIS research. San Francisco. Schwartz, R., Dodge, J., Smith, N., & Etzioni, O. (2019). Green AI. Commun. ACM, 63 (12), 54-63. Shaw, M. (2003). Writing good sotware engineering research papers. 25th International Conerence on Sotware Engineering, (pp. 726-736). Portland. Shea, V. J., Dow, K. E., & Chong, A. (2019). An examination othe long-term business value in investments in inormation technology. Inormation Systems Srontiers: A Journal oResearch and Innovation, 21(1), 213-217. Silver, D., Huang, A., Maddison, C. J., Guez, A., Sire, L., van den Driessche, G., & Hassabis, D. (2016). Mastering the game oGo with deep neural networks and tree search. Nature, (529), 484-489. Sinha, N., Singh, P., Gupta, M., & Singh, P. (2020). Robotics at workplace: An integrated Twitter analytics-SEM based approachor behavioral intention to accept. International Journal oInormation Management, 55, Article 102210. Sipior, J. C. (2020). Considerationsor development and use oAI in response to COVID-19. International Journal oInormation Management, 55, Article 102170. Stone, P., Brooks, R., Brynjolsson, E., Calo, R., Etzioni, O., Hager, G., & Teller, A. (2016). "Artifcial intelligence and lie in 2030. " One hundred year study on artifcial intelligence. Stanord, CA: Stanord University. Sutton, R., & Barto, A. G. (2018). Reinorcement learning: An introduction. MIT Press. The Economist. (2016). Articial Intelligence, The return othe machinery question. The Economist. Trappl, R. (1986). Impacts oarticial intelligence: An overview. In R. Trappl (Ed. ),I. o. Intelligence. Amsterdam: North-Holland. Turing, A. M. (1950). I. —Computing machinery and intelligence. Mind, LIX, 433-460. U. S. National Science and Technology Council. (2016). Preparingor theuture oartifcial intelligence. Washington DC: Government Printing Oce. Vom Brocke, J., Simons, A., Niehaves, B., Riemer, K., Plattaut, R.,&Cleven, A. (2009). Reconstructing the giant: on the importance origour in documenting the literature search process. ECIS. Verona: 17th European Conerence on Inormation Systems (ECIS). von Krogh, G. (2018). Articial intelligence in organizations: New opportunitiesor phenomenon-based theorizing. Academy oManagement Discoveries, 4(4), 404-409. von Neumann, J. (1958). The computer and the brain. New Haven, CT: Yale University Press. Wagner, W. P. (2017). Trends in expert system development: A longitudinal content analysis oover thirty years oexpert system case studies. Expert Systems with Applications, 76, 85-96. Wang, C., Savkin, A., Clout, R., & Nguyen, H. (2015). An intelligent robotic hospital bed or. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 23(5), 744-754. Wang, L., & Wang, X. V. (2016). Outlook ocloud, CPS and Io T in manuacturing. Cloud-Based Cyber-Physical Systems in Manuacturing(pp. 377-398). Cham: Springer. Watson, H. J. (2017). Preparingor the cognitive generation odecision support. MIS Quarterly Executive, 153-169. Webster, J., & Watson, R. (2002). Analyzing the past to prepareor theuture: Writing a literature review. MIS Quarterly, 26(2), xiii-xxiii. Wiener, N. (1948). Cybernetics, or control and communication in the animal and the machine. New York: John Wiley&Sons, Inc. Wilson, J.,&Daugherty, P. R. (2018, July-August). Collaborative Intelligence: Humans and AI Are Joining Forces. Harvard Business Review. Wohlin, C., Runeson, P., H ¨ost, M., Ohlsson, M., Regnell, B., & Wessl ´en, A. (2012). Experimentation in sotware engineering. Springer Science&Business Media. Wole, A. (1991). Mind, sel, society, and computer: Articial intelligence and the sociology omind. American Journal oSociology, 96(5), 1073-1096. Yin, H. H., Langenheldt, K., Harlev, M., Mukkamala, R. R., & Vatrapu, R. (2019). Regulating cryptocurrencies: A supervised machine-learning approach to de-anonymising the bitcoin blockchain. Journal oManagement Inormation Systems, 36, 37-73. Zhang, K.,&Luo, X. (2019). Leveraging Deep-learning and Field Experiment Response Heterogeneity to Enhance Customer Targeting Eectiveness. ICIS 2019 Proceedings. FURTHER READING P1. Meyer, G., Adomavicius, G., Johnson, P. E., Elidrisi, M., Rush, W. A., Sperl-Hillen, J. M., & O'Connor, P. J. (2014). A machine learning approach to improving dynamic decision making Inormation Systems Research, 25, 239-263 P2. Nunamaker, J. E., Derrick, D. C., Elkins, A. C., Burgoon, J. K., & Patton, M. W. (2011). Embodied conversational agent-based kioskor automated interviewing Journal o Management Inormation Systems, 28, 17-48 P3. Gong, J., Abhishek, V., & Li, B. B. (2018). Examining the impact okeyword ambiguity on search advertising perormance: A topic model approach Mis Quarterly, 42, 805 P4. Lukyanenko, R., Parsons, J., Wiersma, Y. F., & Maddah, M. (2019). Expecting the unexpected: Eects odata collection design choices on the quality ocrowdsourced user-generated content Mis Quarterly, 43, 623 P5. Larsen, K. R., Hovorka, D. S., Dennis, A. R., & West, J. D. (2019). Understanding the elephant: The discourse approach to boundary identication and corpus constructionor theory review articles Journal othe Associationor Inormation Systems, 20, 887-927 P6. Aleksander, I. (2017). Partners ohumans: A realistic assessment othe role orobots in theoreseeableuture Journal oInormation Technology, 32, 1-9 P7. Boland, R. J., Jr., & Lyytinen, K. (2017). The limits to language in doing systems design European Journal oInormation Systems, 26, 248-259 P8. Ciborra, C., & Willcocks, L. (2006). The mind or the heart? It depends on the (denition o) situation Journal oInormation Technology, 21, 129-139C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 16P9. Druckenmiller, D. A., & Acar, W. (2009). An agent-based collaborative approach to graphing causal mapsor situationormulation Journal othe Associationor Inormation Systems, 10, 221-251 P10. Markus, M. L. (2017). Datication, organizational strategy, and IS research: What's the score?Journal oStrategic Inormation Systems, 26, 233-241 P11. Mettler, T., Sprenger, M., & Winter, R. (2017). Service robots in hospitals: New perspectives on niche evolution and technology aordances European Journal o Inormation Systems, 26, 451-468 P12. Nissen, M. E., & Sengupta, K. (2006). Incorporating sotware agents into supply chains: Experimental investigation with a procurement task MIS Quarterly: Management Inormation Systems, 30, 145-166 P13. Yin, H. H. S., Langenheldt, K., Harlev, M., Mukkamala, R. R., & Vatrapu, R. (2019). Regulating cryptocurrencies: A supervised machine learning approach to de-anonymizing the bitcoin blockchain Journal oManagement Inormation Systems, 36, 37-73 P14. Aouni, R. (2019). Organizational Learning in the Rise oMachine Learning. ICIS 2019 Proceedings. P15. Baier, L., J ¨ohren, F. &Seebacher, S. (2019). Challenges in the deployment and operation omachine learning in practice. ECIS 2019 Proceedings. P16. Basu, S., Han, W. &Garimella, A. (2019). Impact oArticial Intelligence on Human Decision Making on ICO Platorms. ICIS 2019 Proceedings. P17. Beath, C., Taradar, M. &Ross, J. (2018). One Bank Assure: Customer Intimacy Through Machine Learning. ICIS 2018 Proceedings. P18. Buettner, R., Frick, J. &Rieg, T. (2019). High-perormance detection oepilepsy in seizure-ree EEG recordings: A novel machine learning approach using very specic epileptic EEG sub-bands. ICIS 2019 Proceedings. P19. Chai, Y. &Li, W. (2019). Towards Deep Learning Interpretability: A Topic Modeling Approach. ICIS 2019 Proceedings. P20. Chatterjee, S., Saeedar, P., Toangchi, S. &Kolbe, L. (2018). Intelligent road maintenance: a machine learning approachor surace deect detection. Research Papers. ECIS 2018 Proceedings. P21. Choi, J., Youn, J.-H. &Haas, C. (2019). Machine Learning Approachor Foot-side Classication using a Single Wearable Sensor. ICIS 2019 Proceedings. P22. Dellermann, D., Lipusch, N., Ebel, P. A., Popp, K. M. &Leimeister, J. M. (2017). Finding the Unicorn: Predicting Early Stage Startup Success through a Hybrid Intelligence Method. ICIS 2017 Proceedings. P23. Ghanvatkar, S. &Rajan, V. (2019). Deep Recurrent Neural Networksor Mortality Prediction in Intensive Care using Clinical Time Series at Multiple Resolutions. ICIS 2019 Proceedings. P24. Harouche, A., Quinio, B., Skandrani, S. &Marciniak, R. (2017). A Frameworkor Articial Knowledge Creation in Organizations. ICIS 2017 Proceedings. P25. Homann, P., Oesterle, S., Rust, P. &Urbach, N. (2019). Machine learning approaches along the radiology value chain-rethinking value propositions. ECIS 2019 Proceedings. P26. Keller, R., Stohr, A., Fridgen, G., Lockl, J. &Rieger, A. (2019). Aordance-Experimentation-Actualization Theory in Articial Intelligence Research-A Predictive Maintenance Story. ICIS 2019 Proceedings. P27. Krenzer, A., Stein, N., Griebel, M. &Flath, C. (2019). Augmented Intelligenceor Quality Control oManual Assembly Processes using Industrial Wearable Systems. ICIS 2019 Proceedings. P28. Kuehl, N., Scheurenbrand, J. &Satzger, G. (2016). Needmining: identiying micro blog data containing customer needs. ECIS 2016 Proceedings. P29. Lang, F. &Fink, A. (2014). Decision supportor negotiation protocol selection: a machine learning approach based on articial neural networks. ECIS 2014 Proceedings. P30. Lebovitz, S. (2019). Diagnostic Doubt and Articial Intelligence: An Inductive Field Study oRadiology Work. ICIS 2019 Proceedings. P31. Leyer, M. &Schneider, S. (2019). Me, you or ai? How do weeel about delegation?. ECIS 2019 Proceedings. P32. Liu, X., He, S. &Maedche, A. (2019). Designing an AI-based advisory platormor design techniques. ECIS 2019 Proceedings. P33. Lou, B. &Wu, L. (2019). Articial Intelligence and Drug Innovation. ICIS 2019 Proceedings. P34. Maier, S. B., Jussupow, E. &Heinzl, A. (2019). Good, bad, or both? Measurement o physician's ambivalent attitudes towards AI. ECIS 2019 Proceedings. P35. Maroudas, G. &Louvieris, P. (2006). Exploringactors that determine consumer attitude toward use ointelligent sotware agents. ECIS 2006 Proceedings. P36. Merkert, J., Mueller, M. &Hubl, M. (2015). A Survey othe Application oMachine Learning in Decision Support Systems. ECIS 2015 Proceedings. P37. Meza Martínez, M. A., Nadj, M. &Maedche, A. (2019). Towards an integrative theoreticalramework ointeractive machine learning systems. ECIS 2019 Proceedings. P38. Monu, K. &Woo, C. (2005). Intelligent Agents as a Modeling Paradigm. ICIS 2005 Proceedings. P39. Nagar, Y. &Malone, T. (2011). Making Business Predictions by Combining Human and Machine Intelligence in Prediction Markets. ICIS 2011 Proceedings. P40. Peng, J. (2019). Transer Learning in Dynamic Business Environments: An Application in Earnings Forecastor Public Firms. ICIS 2019 Proceedings. P41. Pumplun, L., Tauchert, C. &Heidt, M. (2019). A new organizational chassisor articial intelligence-exploring organizational readinessactors. ECIS 2019 Proceedings. P42. Ren, J., Ge, H., Wu, X., Wang, G., Wang, W. &Liao, S. (2013). Eective Sentiment Analysis oCorporate Financial Reports. ICIS 2013 Proceedings. P43. Rhue, L. (2019). Beauty's in the AI othe Beholder: How AI Anchors Subjective and Objective Predictions. ICIS 2019 Proceedings. P44. Rzepka, C. &Berger, B. (2018). User Interaction with AI-enabled Systems: A Systematic Review oIS Research. ICIS 2018 Proceedings. P45. Schneider, J. &Handali, J. P. (2019). Personalized explanationor machine learning: a conceptualization. ECIS 2019 Proceedings. P46. Toangchi, S., Hanelt, A. &Li, S. (2019). Advancing Recommendations on Two-Sided Platorms: A Machine Learning Approach to Context-Aware Proling. ICIS 2019 Proceedings. P47. Traumer, F., Oeste-Reiß, S. &Leimeister, J. M. (2017). Towards a Future Reallocation oWork between Humans and Machines-Taxonomy oTasks and Interaction Types in the Context oMachine Learning. ICIS 2017 Proceedings. P48. Urbanke, P., Uhlig, A. &Kranz, J. J. (2017). A Customized and Interpretable Deep Neural Networkor High-Dimensional Business Data-Evidencerom an E-Commerce Application. ICIS 2017 Proceedings. P49. Van Den Broek, E., Sergeeva, A. &Huysman, M. (2019). Hiring Algorithms: An Ethnography oFairness in Practice. ICIS 2019 Proceedings. P50. Wang, H., Huang, J. &Zhang, Z. (2019). The Impact oDeep Learning on Organizational Agility. ICIS 2019 Proceedings. P51. Wang, H., Li, C., Gu, B. &Min, W. (2019). Does AI-based Credit Scoring Improve Financial Inclusion? Evidencerom Online Payday Lending. ICIS 2019 Proceedings. P52. Wang, Q., Liu, X. &Huang, K.-W. (2019). Displaced or Augmented? How does Articial Intelligence Aect Our Jobs: Evidencerom Linked In. ICIS 2019 Proceedings. P53. Weterings, K., Bromuri, S. &Van Eekelen, M. (2019). Explaining customer activation with deep attention model S. ECIS 2019 Proceedings. P54. Wuenderlich, N. V. &Paluch, S. (2017). A Nice and Friendly Chat with a Bot: User Perceptions oAI-Based Service Agents. ICIS 2017 Proceedings. P55. Zhang, K. &Luo, X. (2019). Leveraging Deep-learning and Field Experiment Response Heterogeneity to Enhance Customer Targeting Eectiveness. ICIS 2019 Proceedings. P56. Mahroo, K. (2019). A human-centric perspective exploring the readiness towards smart warehousing: The case oa large retail distribution warehouse International Journal oInormation Management, 45, 176-190 P57. Abubakar, A. M., Behravesh, E., Rezapouraghdam, H., & Yildiz, S. B. (2019). Applying articial intelligence technique to predict knowledge hiding behavior International Journal oInormation Management, 49, 45-57 P58. Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T., & Williams, M. D. (2019). Articial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agendaor research, practice and policy International Journal oInormation Management, Article 101994 P59. Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2019). Articial intelligenceor decision making in the era oBig Data-evolution, challenges and research agenda International Journal oInormation Management, 48, 63-71 P60. Nishant, R., Kennedy, M., & Corbett, J. (2020). Articial intelligenceor sustainability: Challenges, opportunities, and a research agenda International Journal oInormation Management, 53, Article 102104 P61. Sipior, J. C. (2020). Considerationsor development and use oAI in response to COVID-19International Journal oInormation Management, 55, Article 102170 P62. Fan, C., Zhang, C., Yahja, A., & Mostaavi, A. (2019). Disaster City Digital Twin: A visionor integrating articial and human intelligenceor disaster management International Journal oInormation Management, Article 102049 P63. Sung, I., Choi, B., & Nielsen, P. (2020). On the training oa neural networkor online path planning with ofine path planning algorithms International Journal o Inormation Management, Article 102142 P64. Eldrandaly, K. A., Abdel-Basset, M., & Abdel-Fatah, L. (2019). PTZ-Surveillance coverage based on articial intelligenceor smart cities International Journal o Inormation Management, 49, 520-532 P65. Gloor, P., Fronzetti Colladon, A., De Oliveira, J. M., & Rovelli, P. (2020). Put your money where your mouth is: Using deep learning to identiy consumer tribesrom word usage International Journal oInormation Management, 51, Article 101924 P66. Sinha, N., Singh, P., Gupta, M., & Singh, P. (2020). Robotics at workplace: An integrated Twitter analytics-SEM based approachor behavioral intention to accept International Journal oInormation Management, 55, Article 102210 P67. Dwivedi, Y. K., Ismagilova, E., Hughes, D. L., Carlson, J., Filieri, R., Jacobson, J., Jain, V., Karjaluoto, H., Ke, H., Krishen, A. S., Kumar, V., Rahman, M. M., Raman, R., Rauschnabel, P. A., Rowley, J., Salo, J., Tran, G. A., & Wang, Y. (2020). Setting theuture odigital and social media marketing research: Perspectives and research propositions International Journal oInormation Management, Article 102168 P68. Borges, A. F. S., Laurindo, F. J. B., Spínola, M. M., Gonçalves, R. F., & Mattos, C. A. (2020). The strategic use oarticial intelligence in the digital era: Systematic literature review anduture research directions International Journal oInormation Management, Article 102225 P69. Coombs, C. (2020). Will COVID-19 be the tipping pointor the Intelligent Automation owork? A review othe debate and implicationsor research International Journal oInormation Management, 55, Article 102182 P70. Stecher, P., Pohl, M. &Turowski, K. (2020). Enterprise architecture's eects on organizations' ability to adopt articial intelligence-A Resource-based perspective. ECIS 2020 Research Papers. P71. F ¨orster, M., Klier, M., Kluge, K. &Sigler, I. (2020). Evaluating Explainable Artical Intelligence-What Users Really Appreciate. ECIS 2020 Research Papers. P72. Meyer, P., Jonas, J. M. &Roth, A. (2020). Exploring Customers' Acceptance Oand Resistance to Service Robots in Stationary Retail-A Mixed Method Approach. ECIS 2020 Research Papers. P73. Heinrich, K., Gra, J., Chen, J., Laurisch, J. &Zschech, P. (2020). Fool Me Once, Shame On You, Fool Me Twice, Shame On Me: A Taxonomy OAttack And De-Fense Patterns For Ai Security. ECIS 2020 Research Papers. C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
International Journal of Information Management 60 (2021) 102383 17P74. Mehdiyev, N. &Fettke, P. (2020). Prescriptive Process Analytics with Deep Learning and Explainable Articial Intelligence. ECIS 2020 Research Papers. P75. Han, S., Kelly, E., Nikou, S. &Svee, E.-O. (2020). Refections on Articial Intelligence Alignment with Human Values: A Phenomenological Perspective. ECIS 2020 Research Papers. P76. Demlehner, Q. &Laumer, S. (2020). Shall We Use It or Not? Explaining the Adoption oArticial Intelligenceor Car Manuacturing Purposes. ECIS 2020 Research Papers. P77. Dietzmann, C., Heines, R. &Alt, R. (2020). The Convergence oDistributed Ledger Technology and Articial Intelligence: An end-to-end Reerence Lending Processor Financial Services. ECIS 2020 Research Papers. P78. Anton, E., Behne, A. &Teuteberg, F. (2020). The Humans behind Articial Intelligence-An Operationalisation oAI Competencies. ECIS 2020 Research Papers. P79. Sturm, T. &Peters, F. (2020). The Impact oArticial Intelligence on Individual Perormance: Exploring the Fit between Task, Data, and Technology. ECIS 2020 Research Papers. P80. Venumuddala, V. &Kamath, R. (2020). Actualizing aordances-Story oIndian Inormation Technology (IT) Industry delivering Articial Intelligence based solutions. ICIS 2020 Research Papers. P81. Wang, W., Li, B., Luo, X. &Wang, X. (2020). AI Agentsor Sequential Promotions: Combining Deep Reinorcement Learning and Dynamic Field Experimentation. ICIS 2020 Research Papers. P82. Koester, N. &Salge, O. (2020). Building Trust in Intelligent Automation: Insights into Structural Assurance Mechanismsor Autonomous Vehicles. ICIS 2020 Research Papers. P83. Karger, E. (2020). Combining Blockchain and Articial Intelligence-Literature Review and State othe Art. ICIS 2020 Research Papers. P84. Wang, L., Huang, N., Hong, Y., Liu, L., Guo, X. &Chen, G. (2020). Eects oVoice-Based AI in Customer Service: Evidencerom a Natural Experiment. ICIS 2020 Research Papers. P85. F ¨orster, M., Klier, M., Kluge, K. &Sigler, I. (2020). Fostering Human Agency: A Processor the Design oUser-Centric XAI Systems. ICIS 2020 Research Papers. P86. Wanner, J., Heinrich, K., Janiesch, C. &Zschech, P. (2020). How Much AI Do You Require? Decision Factorsor Adopting AI Technology. ICIS 2020 Research Papers. P87. Zapadka, P., Hanelt, A., Firk, S. &Oehmichen, J. (2020). Leveraging“AI-as-a-Service”-Antecedents and Consequences oUsing Articial Intelligence Boundary Resources. ICIS 2020 Research Papers. P88. May, A., Sagodi, A., Dremel, C. &Van Gien, B. (2020). Realizing Digital Innovationrom Articial Intelligence. ICIS 2020 Research Papers. P89. Guo, Y., Yin, X., Liu, D. &Xu, S. X. (2020). “She is not just a computer”: Gender Role oAI Chatbots in Debt Collection. ICIS 2020 Research Papers. P90. Deng, Y., Zheng, J., Huang, L. &Kannan, K. (2020). The Impact oAI-powered Shel Monitoring on Product Sales. ICIS 2020 Research Papers. P91. Ochmann, J., Michels, L., Zilker, S., Tieenbeck, V. &Laumer, S. (2020). The infuence oalgorithm aversion and anthropomorphic agent design on the acceptance oAI-based job recommendations. ICIS 2020 Research Papers. P92. Savarimuthu, B. T. R., Corbett, J., Yasir, M. &Lakshmi, V. (2020). Using Machine Learning to Improve the Sustainability othe Online Review Market. ICIS 2020 Research Papers. P93. Grønsund, T., & Aanestad, M. (2020). Augmenting the algorithm: Emerging human-in-the-loop work congurations The Journal oStrategic Inormation Systems, 29, Article 101614 P94. Zhu, H., Samtani, S., Chen, H., & Nunamaker, J. F., Jr. (2020). Human identication or activities odaily living: A deep transer learning approach Journal o Management Inormation Systems, 37, 457-483 P95. Smith, G. (2020). Data miningool's gold Journal oInormation Technology, 35, 182-194 P96. Schuetz, S., & Venkatesh, V. (2020). Research perspectives: The rise ohuman machines: How cognitive computing systems challenge assumptions ouser-system interaction Journal othe Associationor Inormation Systems, 21, 460-482 P97. Coombs, C., Hislop, D., Taneva, S. K., & Barnard, S. (2020). The strategic impacts o Intelligent Automationor knowledge and service work: An interdisciplinary review The Journal oStrategic Inormation Systems, 29, Article 101600 P98. Mostaa, M. M., & El-Masry, A. A. (2013). Citizens as consumers: Proling e-government services' users in Egypt via data mining techniques International Journal oInormation Management, 33, 627-641C. Collins et al. | Artificial intelligence in information systems research_ A systematic literature review and research agenda - 1-s2.0-S0268401221000761-main.pdf |
README.md exists but content is empty.
Use the Edit dataset card button to edit it.
- Downloads last month
- 0