AIS Logo

Got a research question instead? Ask AIS Insights

Browse the library 79 papers

All Artificial Intelligence & Automation Information Systems & Management Human-Computer Interaction Data Analytics & Big Data Digital Transformation & Strategy Cybersecurity & Privacy Cloud & Infrastructure E-Commerce & Digital Platforms Operations & Supply Chain Sustainability & Green IS
Artificial Normality: How Conversational Agents’ Perceived Humanness Inhibits Error Attribution and Preserves Satisfaction
AI Processed Human Approved

How does making conversational agents appear more human affect customer reactions when errors occur?

Alfred Benedikt Brendel, Sascha Lichtenberg, Fabian Hildebrandt & Alan R. Dennis
Artificial Normality: How Conversational Agents’ Perceived Humanness Inhibits Error Attribution and Preserves Satisfaction
Journal of Management Information Systems · 2026
Key findings
  • Greater perceived humanness in conversational agents preserves perceptions of situational normality when minor errors occur.
  • Preserved situational normality reduces users' tendency to attribute error cause to the agent, thereby safeguarding overall service satisfaction.
  • Errors made by humanlike conversational agents are processed similarly to errors made by human service agents, demonstrating that humanlike design can alleviate algorithm aversion.
What it means for you
  • CIO / IT Executive: On Monday morning, allocate budget to pilot the integration of more human-centric design principles (e.g., nuanced language, empathetic phrasing) into our core customer-facing conversational agent, focusing on areas where process errors are most common, to preemptively reduce user error attribution and maintain satisfaction.
  • IT Manager: On Monday morning, task your development team to review our current conversational agent's dialogue logs for instances of process errors and identify the top 3 most frequent error types to begin redesigning their responses with more humanlike cues (e.g., slight pauses, conversational fillers) to normalize the situation for users.
  • Business Strategist: On Monday morning, initiate a cross-functional workshop with marketing and product development to explore how to strategically position our conversational agents' perceived humanness in customer communications, emphasizing its role in creating a seamless experience even when minor technical glitches occur.
  • Researcher: On Monday morning, begin designing a follow-up study that experimentally manipulates specific humanlike conversational features (e.g., varying levels of politeness, acknowledgment of difficulty) within a simulated error scenario to quantify their direct impact on error attribution and situational normality, building upon the findings of Artificial Normality.
  • Policymaker: On Monday morning, convene a small working group to draft guidelines for the ethical development of conversational agents, recommending that AI design standards encourage the incorporation of humanlike qualities to enhance user experience and mitigate negative reactions to inevitable system errors, rather than solely focusing on pure efficiency.
Chatbot, Chatbot errors, situational normality, anthropomorphism, humanness, error attribution, user satisfaction
What Does ChatGPT Know About Information Systems?
AI Processed Human Approved

How well does ChatGPT perform on information systems course questions and professional certification exams?

Daniel E. O’Leary, Veda C. Storey, Aaron M. French, Joseph R. Buckman, Cecil Chua, Andrew W. Green, Grace Gu, Fred Niederman, Francis Pereira, Gary Templeton, Linda Wallace
What Does ChatGPT Know About Information Systems?
Communications of the Association for Information Systems · 2025
Key findings
  • ChatGPT answered 65% to 85% of Information Systems queries correctly across all core framework areas, performing roughly at the level of an average student.
  • The model performed significantly better on university course exams than on professional certification exams, failing to reach passing scores for the PMP and Microsoft Azure Administrator tests.
  • ChatGPT achieved higher relative accuracy against class averages on closed-book exams compared to open-book or online exam environments where students had external resource access.
  • While ChatGPT displayed broad coverage across diverse IS topics, its responses favored listing structured information over demonstrating expert-level critical thinking or complex problem-solving.
What it means for you
  • CIO / IT Executive: On Monday morning, task your IT training lead to pilot a small internal workshop for a select group of IT professionals focusing on the PMP and Microsoft Azure Administrator certifications, using ChatGPT as a supplementary learning tool to identify specific knowledge gaps and assess its effectiveness in preparing for these professional exams.
  • IT Manager: On Monday morning, instruct your team members responsible for generating internal documentation or knowledge base articles to compare the accuracy and completeness of ChatGPT's initial drafts on at least two core Information Systems topics (e.g., database management, network security) against your existing, verified internal resources.
  • Business Strategist: On Monday morning, task a market analyst to explore potential use cases for ChatGPT in rapidly generating initial drafts of business process documentation or competitive analysis reports within your IS domain, noting its tendency to provide structured information over deep critical thinking.
  • Researcher: On Monday morning, design a controlled experiment to compare ChatGPT's performance on retrieving specific, factual information about IS concepts in a closed-book scenario versus an open-book scenario using publicly available academic papers, to validate the finding about its performance in different exam environments.
  • Policymaker: On Monday morning, request a briefing from your education and technology policy advisors on the implications of AI tools like ChatGPT achieving 'average student' performance in IS education, and explore potential guidelines for incorporating such tools responsibly in curriculum development and assessment.
ChatGPT, Large Language Models, Generative AI, Information Systems Competency, Artificial Intelligence in Education, Knowledge Assessment
GENAI AS A ROMANTIC RELATIONSHIP ADVISOR: EXPLORING HUMAN–AI COLLABORATION IN EMOTIONAL SUPPORT CONTEXTS
AI Processed Human Approved

What are the benefits and risks of using generative artificial intelligence for romantic relationship advice?

Anna Schätzle, Peter Buxmann
GENAI AS A ROMANTIC RELATIONSHIP ADVISOR: EXPLORING HUMAN–AI COLLABORATION IN EMOTIONAL SUPPORT CONTEXTS
ECIS 2026 · 2026
Key findings
  • Qualitative evidence supported that GenAI helps users self-clarify emotions, organize thoughts, and prepare communication strategies prior to interpersonal discussions.
  • Qualitative evidence supported that GenAI use entails risks of automation and confirmation bias, as the system tends to validate the user's one-sided, potentially biased framing of relationship conflicts.
  • Qualitative evidence supported that users often attribute unintended authority to GenAI, potentially misinterpreting its structured, polished responses as clinical diagnoses.
  • Qualitative analysis supported that GenAI is a useful complementary or preparatory tool but cannot replace professional therapy because it lacks emotional intuition, nonverbal communication perception, and dyadic mediation capabilities.
  • No statistical hypotheses were tested due to the study's exploratory qualitative nature; findings are limited by a small sample size of 18 interviewees, retrospective self-reports, a German-speaking context, and homogeneous use of ChatGPT.
What it means for you
  • CIO / IT Executive: Configure the enterprise GenAI gateway to detect personal counseling or emotional-support-seeking queries and automatically inject a system-level disclaimer redirecting employees to the company's official Employee Assistance Program (EAP) to prevent reliance on GenAI as an authoritative clinical resource.
  • IT Manager: Create and distribute a 'Balanced Prompting' template to the team, instructing them to explicitly prompt GenAI to 'act as a devil's advocate and challenge my assumptions' when using it to prepare for difficult workplace or interpersonal conversations, directly mitigating the tool's confirmation bias.
  • Business Strategist: Review the product roadmap for our wellness and conversational AI features to pivot the UX away from direct 'advice-giving' toward 'communication preparation and self-reflection' tools, integrating mandatory disclaimers that clarify the AI has no clinical diagnostic capabilities.
  • Researcher: Draft a research proposal for a pre-registered quantitative study ($N > 300$) comparing how different LLM models (e.g., ChatGPT, Claude, Gemini) handle relationship conflict prompts, specifically measuring the rate of user agreement (automation bias) versus the model's level of validation.
  • Policymaker: Draft a regulatory policy mandate requiring developers of consumer-facing GenAI tools to implement prominent 'non-clinical tool' labels and active-consent pop-ups warning users of the risks of automation bias and the lack of professional diagnostic authority in emotionally sensitive domains.
Generative Artificial Intelligence, Relationship, Counseling, Therapy, Human-AI Interaction, HAII-TIME
MANAGING ARTIFICIAL INTELLIGENCE IN PUBLIC SECTOR ORGANIZATIONS: A DYNAMIC CAPABILITIES PERSPECTIVE
AI Processed Human Approved

How can public sector organizations effectively adopt and manage artificial intelligence?

Isabella Urban, Manfred Baer, Ralf Plattfaut
MANAGING ARTIFICIAL INTELLIGENCE IN PUBLIC SECTOR ORGANIZATIONS: A DYNAMIC CAPABILITIES PERSPECTIVE
ECIS 2026 · 2026
Key findings
  • The study empirically supports that AI adoption in public organizations is hindered by high levels of technological, regulatory, and organizational uncertainties, which require proactive management.
  • The research empirically identifies and supports a framework of specific organizational microfoundations across three key dimensions: sensing (e.g., tracing employee expectations and monitoring regulatory landscapes), seizing (e.g., establishing AI governance and launching pilot projects), and transforming (e.g., active change management and promoting AI literacy).
  • The findings empirically show that employee representation bodies (such as staff councils) often play a critical, gatekeeping role that can delay or hinder AI implementation decisions.
  • The study acknowledges limitations, including its geographical focus on the German public sector and a limited observation timeframe from April 2024 to October 2025.
What it means for you
  • CIO / IT Executive: Send an email calendar invite to the chair of your organization's Employee Representation Body (Staff Council) to establish a joint, bi-weekly 'AI Governance and Alignment Committee' to co-create deployment guidelines and prevent approval bottlenecks.
  • IT Manager: Draft a 1-page project charter for a highly constrained, 4-week pilot project using a low-risk internal AI tool (e.g., automated document summarizing) to test technical capability and establish baseline performance metrics.
  • Business Strategist: Design and launch a 5-question MS Forms survey to all department heads to map their teams' current AI literacy levels and identify specific anxieties, using the results to structure next month's targeted change management plan.
  • Researcher: Draft a research proposal and reach out to two international public sector research partners to replicate this qualitative dynamic capabilities study in a non-German jurisdiction (e.g., the US or UK) to address the geographical limitations of the original study.
  • Policymaker: Create and distribute a standardized 'Public Sector AI Regulatory Compliance Checklist' to provide local agency leaders with clear, step-by-step guidance on data privacy and municipal regulations, directly reducing their regulatory uncertainty.
Artificial Intelligence, Dynamic Capabilities, Microfoundations, Multiple Case Study, Public Sector
TIME TO TURN OFF THE LIGHTS: A SUCCESS FACTOR FRAMEWORK FOR ADVANCING HYPERAUTOMATION IN ORGANIZATIONS
AI Processed Human Approved

What are the key factors needed to successfully scale automation across an organization?

Lisa Straub, Christian Zeiß, Ulvi Ibrahimli, Christoph Tomitza, Benedikt Wirsing, Axel Winkelmann
TIME TO TURN OFF THE LIGHTS: A SUCCESS FACTOR FRAMEWORK FOR ADVANCING HYPERAUTOMATION IN ORGANIZATIONS
ECIS 2026 · 2026
Key findings
  • The survey results statistically supported all 15 success factors within the framework, achieving a high average agreement score of 6.05 out of 7, with 93.2% of respondents selecting positive agreement.
  • 'Leadership Commitment' (SF3) and 'Communication and Trust Building' (SF4) received the highest levels of overall agreement, indicating strong statistical support.
  • 'Strategic Partner Collaboration' (SF15) and 'Customer Integration' (SF13) received statistically supported relevance, although they had comparatively lower overall agreement scores.
  • The study's generalizability is limited by the German context of the initial qualitative interviews, a modest quantitative survey sample size (n = 103), and its reliance on subjective assessments rather than objective performance metrics.
What it means for you
  • CIO / IT Executive: On Monday morning, schedule a 30-minute alignment meeting with business unit heads to establish a joint 'Hyperautomation Governance Board' to audit all active shadow-IT automation tools (like RPA and low-code apps) and map them against your long-term IT integration roadmap to prevent fragmented technical debt.
  • IT Manager: On Monday morning, block out two hours to audit your current automation deployment backlog and draft a standardized 'API-First Integration Checklist' that your development team must use to ensure new automations do not bypass APIs and create system incompatibilities.
  • Business Strategist: On Monday morning, select one high-volume customer-facing workflow and schedule a scoping call with a key strategic partner to co-design an integrated, automated data-sharing process, directly applying the 'Customer Integration' (SF13) and 'Strategic Partner Collaboration' (SF15) success factors.
  • Researcher: On Monday morning, draft a research proposal and outreach email to target three non-German, US-based enterprises to secure pilot sites for a comparative case study that measures objective operational KPIs (such as transaction error rates and processing speed) to validate the subjective findings of the 15-factor framework.
  • Policymaker: On Monday morning, draft an amendment to your organization's IT Procurement Policy mandating that any new business unit purchasing low-code or RPA tools must submit a 'Systemic Compatibility and Trust Impact Assessment' addressing how the tool aligns with the enterprise IT architecture and employee trust-building guidelines.
Hyperautomation, Transformation Management, Organizational Change, Success Factor Framework, Affordance Theory
ARE THE REGULATORS WINNING THE FIGHT WITH ANONYMITY-ENHANCING COINS? STUDYING BINANCE’S EUROPEAN DELISTING OF MONERO AS A NATURAL EXPERIMENT
AI Processed Human Approved

How does delisting a privacy coin from a major exchange impact its price over time?

Robin Renwick, Rob Gleasure
ARE THE REGULATORS WINNING THE FIGHT WITH ANONYMITY-ENHANCING COINS? STUDYING BINANCE’S EUROPEAN DELISTING OF MONERO AS A NATURAL EXPERIMENT
ECIS 2026 · 2026
Key findings
  • Hypothesis 1 (delisting will cause an immediate negative price impact) was STATISTICALLY SUPPORTED: Monero's price was significantly lower (estimated at 63% to 72% lower) post-delisting compared to its synthetic counterfactual.\n- Hypothesis 2 (the negative price impact will diminish over time) was NOT SUPPORTED: Rather than recovering, the negative price impact intensified when longer time lags of up to 200 days were examined.\n- Limitations: The findings are restricted to a single delisting event of one specific privacy coin (Monero) and use trading price as an imperfect proxy for perceived value.
What it means for you
  • CIO / IT Executive: Audit our organization's digital asset holdings and treasury to identify any exposure to privacy coins like Monero, and immediately initiate a migration plan to convert these assets to compliant, liquid cryptocurrencies to avoid permanent valuation loss from regulatory delistings.
  • IT Manager: Update network monitoring tools and firewall rules to detect and block any Monero nodes, mining scripts, or wallet connections on company-owned infrastructure to prevent compliance violations arising from tightening centralized exchange regulations.
  • Business Strategist: Review our payment gateway integrations and product roadmap to remove Monero or other privacy coins as accepted payment methods, and pivot our privacy-focused initiatives toward compliant layer-2 scaling solutions (like Zero-Knowledge rollups) that satisfy regulators.
  • Researcher: Draft a new research proposal to apply the synthetic control method to other privacy-centric assets (like Zcash or Dash) across different geographic regions, incorporating decentralized exchange (DEX) trading volumes to test if the value shifted off centralized order books.
  • Policymaker: Draft an amendment to the national crypto-asset regulatory framework to mandate that all registered centralized exchanges delist anonymity-enhancing coins, citing this empirical evidence that exchange-level bans are highly effective at permanently reducing these assets' market value and liquidity.
Cryptocurrencies, Monero, Privacy, Regulation, Natural Experiment, Synthetic Control Method
HARNESSING AND REDUCING THE GENERATIVE AND WORKFLOW AUTONOMY OF GENERATIVE AI SYSTEMS: A MULTIPLE-CASE STUDY
AI Processed Human Approved

How can businesses manage generative AI autonomy after deployment to reduce risks without losing value?

Vincent Heimburg, Manuel Wiesche
HARNESSING AND REDUCING THE GENERATIVE AND WORKFLOW AUTONOMY OF GENERATIVE AI SYSTEMS: A MULTIPLE-CASE STUDY
ECIS 2026 · 2026
Key findings
  • The study conceptualizes AI autonomy as consisting of two dimensions: generative autonomy (content creation) and workflow autonomy (process and system interaction), which is qualitatively supported by the case analysis.
  • The research identifies 20 distinct post-deployment practices categorized under either harnessing value-adding autonomy (such as real-time auditing and semantic resource enrichment) or reducing unproductive autonomy (such as architectural weaving and resource circuit breakers), which is qualitatively supported by the case analysis.
  • The analysis identifies 'averting effects' where efforts to manage one dimension of autonomy require maintaining the other, which is qualitatively supported by the case analysis.
  • The study identifies a 'potential' relationship where reducing workflow autonomy creates a structural foundation that allows providers to further restrict generative autonomy, which is qualitatively supported by the case analysis.
  • The findings are limited by the exploratory, qualitative nature of the seven cases, which does not quantitatively measure the operational efficiency, safety, or economic impacts of the individual practices, as well as the limited temporal autonomy of current systems.
What it means for you
  • CIO / IT Executive: On Monday morning, audit your organization's active GenAI integrations and mandate the immediate implementation of 'resource circuit breakers'—hard programmatic limits on API calls and database writes—to restrict workflow autonomy and prevent runaway system actions.
  • IT Manager: On Monday morning, set up a 'real-time auditing' log pipeline for your team's customer-facing GenAI application to flag and review outputs where the system's confidence score drops below 80%, immediately routing those queries to human-in-the-loop agents.
  • Business Strategist: On Monday morning, map your company's GenAI product features onto a 2x2 grid of Generative Autonomy vs. Workflow Autonomy to identify features where you can reduce risky workflow autonomy (e.g., auto-sending emails) without degrading the value of the generated content.
  • Researcher: On Monday morning, draft a research proposal for a quantitative empirical study that measures the operational latency, safety rates, and economic impacts of 'architectural weaving' vs. 'semantic resource enrichment' across B2B GenAI deployments.
  • Policymaker: On Monday morning, draft an update to your agency's AI procurement guidelines requiring vendors to prove they have 'averting effects' controls in place, specifically demonstrating how their systems constrain generative output risks when workflow autonomy is elevated.
Artificial intelligence, GenAI, Agentic AI, Large Language Models, Case Study, Autonomy
STRUCTURING READINESS FOR AI INNOVATION: THE ROLE OF THE DESIGN CHALLENGE BRIEF AS A COORDINATION AND GATEKEEPING ARTEFACT
AI Processed Human Approved

How can teams evaluate whether an artificial intelligence project is ready to launch?

Jennifer Hehn, Franziska Anna Röckel, Benjamin van Giffen
STRUCTURING READINESS FOR AI INNOVATION: THE ROLE OF THE DESIGN CHALLENGE BRIEF AS A COORDINATION AND GATEKEEPING ARTEFACT
ECIS 2026 · 2026
Key findings
  • The study outlines a preliminary three-phase qualitative model for establishing AI project readiness: divergent exploration, negotiation/translation, and convergent alignment, though these findings are descriptive and lack statistical testing.
  • The Design Challenge Brief is shown to act as a gatekeeping artifact that structures readiness reviews and allows teams to selectively revisit earlier project assumptions without resetting the entire process.
  • The study is limited by its small qualitative sample of five projects and its setting within a structured, university-led innovation lab, which may restrict generalizability to less-structured corporate settings.
What it means for you
  • CIO / IT Executive: Draft and mandate a standard 'Design Challenge Brief' template that all AI project champions must complete and get approved before receiving formal project launch funding.
  • IT Manager: Gather your AI project team for a 1-hour workshop to map your current pre-project planning stage to the three phases of divergent exploration, negotiation/translation, and convergent alignment, explicitly documenting how you will handle fluid data availability.
  • Business Strategist: Audit your pipeline of proposed AI initiatives and flag any project that has not yet completed a collaborative 'Design Challenge Brief' to prevent launching projects with high uncertainty around non-deterministic outcomes.
  • Researcher: Draft a research proposal and survey instrument to test the generalizability of the three-phase AI readiness model across a quantitative sample of 100+ corporate AI projects outside of academic lab settings.
  • Policymaker: Integrate a 'Design Challenge Brief' gatekeeping requirement into your public sector AI procurement and funding frameworks to ensure public agencies evaluate data readiness and outcome uncertainties before awarding AI contracts.
AI innovation, readiness work, design challenge brief, boundary objects, gatekeeping
SCHNITZEL-PREDICTION: DESIGNING HUMAN–AI COLLABORATION FOR CAFETERIA DEMAND FORECASTING
AI Processed

How does artificial intelligence compare to experienced human planners when forecasting cafeteria meal demand?

Justus Cappel, Timo Strohmann, Mara Burger, Marleen Voß, Jan vom Brocke
SCHNITZEL-PREDICTION: DESIGNING HUMAN–AI COLLABORATION FOR CAFETERIA DEMAND FORECASTING
ECIS 2026 · 2026
Key findings
  • The algorithmic machine learning model achieved a 30% reduction in forecasting errors compared to a naive 7-day lag baseline.
  • The hypothesis that the algorithmic model would statistically outperform experienced human planners across all menu items was not supported, as direct statistical comparisons showed no overall significant difference (p = 0.454).
  • Experienced human planners statistically outperformed the algorithmic model for novel menu items with limited historical data (MAE of 10.46 for humans versus 17.00 for the algorithm), supporting the hypothesis that human judgment remains superior for high-uncertainty contextual forecasting.
  • Acknowledgeable limitations of this research include its single-case design at a German firm, a small quantitative evaluation scope of 24 operating days, and qualitative insights derived from two think-aloud sessions.
What it means for you
  • CIO / IT Executive: Issue a directive to your enterprise architecture team to halt all pure-automation forecasting initiatives and mandate that any future demand-planning software deployments integrate a 'human-in-the-loop' interface, specifically reserving override capabilities for experienced frontline staff in high-uncertainty scenarios.
  • IT Manager: Create a task in your team's active development sprint to build an LLM-powered semantic tagging feature for your menu database, and pair it with a simple UI override slider on the kitchen manager's dashboard to let them manually adjust AI-generated forecasts for newly introduced dishes.
  • Business Strategist: Review your cafeteria's historical food waste logs and restructure the kitchen managers' weekly KPIs to incentivize them to manually audit and adjust automated algorithmic forecasts, focusing their human intervention specifically on novel menu items where historical data is sparse.
  • Researcher: Draft a grant proposal for a multi-site, longitudinal study that replicates the 'SCHNITZEL-PREDICTION' methodology across five different corporate catering firms over a six-month period to overcome the single-case and 24-day evaluation limitations of this study.
  • Policymaker: Draft an update to the public sector sustainable procurement guidelines requiring that any government-funded cafeteria software tenders mandate human-AI collaborative interfaces rather than fully automated solutions to actively prevent food waste caused by algorithmic 'cold-start' forecasting failures.
Human-AI Collaboration, Action Design Research, Sustainable IS, Demand Forecasting, Food Waste, Large Language Models
HOW GENERATIVE AI DRIVES FIELD-LEVEL CHANGE IN MANAGEMENT CONSULTING
AI Processed

How does using generative AI in everyday work change how management consulting firms operate and deliver value?

Robin Killewald, Thomas Haskamp
HOW GENERATIVE AI DRIVES FIELD-LEVEL CHANGE IN MANAGEMENT CONSULTING
ECIS 2026 · 2026
Key findings
  • The study identified four core mechanisms of change—negotiated legitimacy, provisionality of reliance, skill polarity, and contestation of value—which were qualitatively supported by empirical interview data.
  • Under the 'provisionality of reliance' mechanism, temporal pressures cause consultants to bypass validation routines, creating fragile outputs that risk client trust (supported qualitatively).
  • Under the 'skill polarity' mechanism, generative AI simultaneously expands short-term capabilities and erodes long-term analytical thinking and mentoring routines (supported qualitatively).
  • Under the 'contestation of value' mechanism, the commoditization of standardized tasks shifts the focus of consulting value toward human-centered contextual judgment and experience (supported qualitatively).
  • Quantitative hypothesis testing was not performed due to the study's qualitative, inductive methodology, which limits statistical generalizability to other contexts or regions.
What it means for you
  • CIO / IT Executive: Integrate an automated 'Validation Gateway' into the firm’s proprietary GenAI workspace that blocks users from exporting drafts into client-ready formats (like PPTX or PDF) until they complete a digital checklist verifying the accuracy and sources of the underlying data.
  • IT Manager: Set up a weekly automated audit report in your GenAI usage analytics platform to identify teams with high-volume AI prompt generations but low average document-editing times, and flag these teams for manual quality-assurance reviews.
  • Business Strategist: Redesign the firm's standard client engagement proposal template to phase out hourly billing for commoditized deliverables (like market research summaries) and replace it with value-based pricing focused on human-led strategic judgment workshops.
  • Researcher: Draft a quantitative survey instrument targeting 500 management consultants across North America and Asia to test if the qualitative mechanisms of 'skill polarity' and 'provisionality of reliance' negatively correlate with long-term client retention metrics.
  • Policymaker: Introduce an industry-standard certification framework for professional service firms that mandates documented 'human-in-the-loop' validation protocols and minimum structured mentoring hours to protect junior workforce development against generative AI skill erosion.
Generative AI, Professional Service Firms, Causal Mechanisms, Routines, Management Consulting
RETHINKING KNOWLEDGE WORK: DESIGNING LLM-BASED SYSTEMS FOR COMPLEXITY MANAGEMENT
AI Processed Human Approved

How effectively can artificial intelligence tools screen and extract information from complex procurement documents?

Moritz Diener, Simon Kaps, Philipp Spitzer, Robin Hirt, Michael Vössing, Gerhard Satzger
RETHINKING KNOWLEDGE WORK: DESIGNING LLM-BASED SYSTEMS FOR COMPLEXITY MANAGEMENT
ECIS 2026 · 2026
Key findings
  • The LLM-based system reduced initial document screening time by 59% (from an average of 3 minutes 42 seconds manually to 1 minute 30 seconds via the system), which was statistically supported.
  • The system achieved an overall content-match accuracy of 88.82% for text extraction, with highest accuracy for identifying the procuring entity and place of execution (94.10% each) and lowest for the type of service (79.40%).
  • For binary classification tasks, the system achieved an average accuracy of 83.85%, performing significantly better on detecting required certifications (91.20%) than on required references (76.50%).
  • Qualitative feedback supported the system's ability to lower perceived document, compliance, and process complexity, shifting human workflows from manual reading and extraction to output verification.
  • Limitations include testing within a single German SME in the data center sector, a short-term study duration, and a small evaluation dataset (34 documents), restricting direct generalizability to other industries.
What it means for you
  • CIO / IT Executive: On Monday morning, issue an IT directive mandating that any generative AI tools deployed for document screening must include a 'human-in-the-loop' verification interface, explicitly prohibiting automated final submissions due to the system's average 11.18% error rate in text extraction.
  • IT Manager: On Monday morning, configure your document-processing pipeline's UI to visually highlight and force manual validation of 'type of service' and 'required references' fields, as the research shows these have lower extraction accuracies (79.4% and 76.5%) compared to highly reliable fields like 'procuring entity' (94.1%).
  • Business Strategist: On Monday morning, calculate the potential ROI of deploying an LLM-based screening tool for your proposal-writing team, using the study's proven 59% reduction in initial document screening time (down to 1 minute 30 seconds per document) to project resource hours saved.
  • Researcher: On Monday morning, write a research proposal to extend this design science framework by testing it on a dataset larger than 34 documents across multi-industry and non-German settings, focusing specifically on optimizing prompts to improve the low 76.5% accuracy baseline for identifying required references.
  • Policymaker: On Monday morning, initiate a draft proposal for public procurement formatting standards that mandates standardized, machine-readable sections for 'required certifications' and 'references' to minimize AI extraction errors (currently at 8.8% and 23.5% respectively) and ensure fair bidding processes for SMEs.
large language models, generative AI, complexity management, knowledge work, design science research
DESIGNING FOR IMPERFECT PROMPTS: A SIMULATION-BASED EVALUATION OF AN LLM-BASED ENTERPRISE DATA ASSISTANT USING KNOWLEDGE GRAPHS
AI Processed Human Approved

Which approach best helps AI assistants translate imperfect user prompts into accurate database queries?

Niklas Wagner, Alexander Mdche
DESIGNING FOR IMPERFECT PROMPTS: A SIMULATION-BASED EVALUATION OF AN LLM-BASED ENTERPRISE DATA ASSISTANT USING KNOWLEDGE GRAPHS
ECIS 2026 · 2026
Key findings
  • The KG-based mediation statistically outperformed the schema-based mediation across Precision, Recall, and F1-Score metrics.\n- For GPT-4.1, the KG-based approach achieved a Precision of 90.67% (compared to 69.05% for schema-based) and an F1-Score of 89.47% (compared to 71.02% for schema-based).\n- For Claude Sonnet 4, the KG-based approach achieved a Precision of 87.23% (compared to 75.17% for schema-based) and an F1-Score of 90.29% (compared to 82.17% for schema-based), while Recall remained statistically comparable (93.52% vs 92.56%).\n- The study is limited by its focus on a single-enterprise setting within a CFO unit, a finite dataset of 158 interactions from 29 users, and reliance on a simulation-based evaluation instead of live field deployments.
What it means for you
  • CIO / IT Executive: Issue a directive to transition the enterprise's LLM-based data assistant architecture away from simple schema-based semantic similarity matching and onto a Knowledge Graph-based mediation layer constructed from database schemas to capture the demonstrated 12% to 21% precision improvements.
  • IT Manager: Instruct your database and engineering teams to begin constructing a Knowledge Graph from your current database schemas to serve as a semantic mediation layer for your LLM assistant, replacing manual prompt engineering.
  • Business Strategist: Launch a pilot program within the CFO business unit to deploy the Knowledge Graph-based data assistant, leveraging its verified ability to resolve incomplete, domain-specific prompts into accurate database queries for business users.
  • Researcher: Draft a research proposal to test the Knowledge Graph-based mediation layer in a live, real-world field deployment to gather empirical performance data beyond the current study's simulation of 158 interactions across 29 users.
  • Policymaker: Draft an internal data governance policy requiring that all natural-language LLM interfaces querying corporate databases must use a schema-derived Knowledge Graph mediation layer to ensure structural query alignment and prevent inaccurate data retrieval.
Large Language Models, Enterprise Data Assistants, Text-to-SQL, Knowledge Graphs, Design Science Research
FROM BROWN TO GREEN: HOW AI-DRIVEN TRANSFORMATION ENABLES BUSINESS MODEL RECONFIGURATION IN THE TWIN TRANSITION
AI Processed Human Approved

How can businesses use artificial intelligence to transition their business models toward sustainability?

Elaine Mosconi, Marie-Claude Boudreau
FROM BROWN TO GREEN: HOW AI-DRIVEN TRANSFORMATION ENABLES BUSINESS MODEL RECONFIGURATION IN THE TWIN TRANSITION
ECIS 2026 · 2026
Key findings
  • The AI-enabled platform qualitatively supported a business model shift from transactional, time-based billing to recurring, outcome-based services focused on asset durability and lifecycle performance.
  • A strategic alignment gap was identified: executives framed the AI and sustainability transformation as radical, whereas operational managers experienced it as incremental.
  • Successful scaling of the AI initiative qualitatively depended on adaptive governance (looser decision-making, rapid experimentation), pre-requisite capability building in data engineering, and strong leadership support.
  • Key organizational trade-offs and barriers emerged, specifically related to skills gaps, legacy system integration, high energy consumption of AI models, and delayed financial return on investment.
  • Limitation: The findings of this study are based on a single qualitative case study of a Canadian engineering firm, which may limit the direct generalizability of the results to other industries or regions.
What it means for you
  • CIO / IT Executive: Initiate an audit of legacy data pipelines to identify bottlenecks for AI integration, and establish a 'green sandbox' environment with relaxed governance rules to allow developers to experiment with asset-durability algorithms while tracking model energy consumption.
  • IT Manager: Hold a 30-minute team sync to map out existing data engineering skill gaps and assign one developer to clean and structure a single historical dataset of asset lifecycle performance to prepare it for future AI model training.
  • Business Strategist: Draft a pilot proposal to transition one high-trust client from hourly billing to a recurring, outcome-based pricing model focused on extending their asset lifecycle, and schedule a meeting with operational managers to align on realistic, incremental milestones.
  • Researcher: Design a survey instrument targeting business and IT leaders across multiple industries to quantitatively test whether the perceptual gap between executive (radical) and operational (incremental) views on AI-driven green transitions exists outside of the engineering sector.
  • Policymaker: Draft a policy brief proposing a targeted grant program that subsidizes data-engineering upskilling and offsets energy-efficiency auditing costs for mid-sized firms adopting AI-enabled, sustainable business models.
Twin Transition, Artificial Intelligence, Sustainability, Business Model, Dynamic Capabilities
THE AI-ENABLED CIRCULAR ECONOMY: EXTRACTING AND SYNTHESIZING DESIGN KNOWLEDGE
AI Processed

How can organizations design and use artificial intelligence to support circular economy practices?

Linda Sagnier Eckert, Christoph Hoppe-Ludwig, Timo Strohmann, Anne Ixmeier, Daniel Heinz
THE AI-ENABLED CIRCULAR ECONOMY: EXTRACTING AND SYNTHESIZING DESIGN KNOWLEDGE
ECIS 2026 · 2026
Key findings
  • Synthesized design knowledge from 99 empirically grounded studies supported four distinct meta-principles for AI-enabled circular economies.
  • Meta-Principle 1 (AI for slowing resource loops) is supported by 15 empirical studies showing predictive AI models successfully forecast asset degradation to trigger life-extending actions.
  • Meta-Principle 2 (AI for closing resource loops) is supported by 29 empirical studies showing integrated computer vision and robotic sorting systems reliably classify and separate materials.
  • Meta-Principle 3 (AI for narrowing resource loops) is supported by 18 empirical studies showing hybrid optimization models successfully reduce resource and energy consumption.
  • Meta-Principle 4 (AI for socio-technical orchestration) is supported by 37 empirical studies demonstrating that AI can drive the organizational dynamic capabilities needed to coordinate circular workflows.
  • A recognized limitation is that the corpus is restricted to English-language, peer-reviewed publications from the last five years, and the synthesized principles have not been longitudinally tested in new field environments.
What it means for you
  • CIO / IT Executive: Schedule a 9:00 AM workshop with your Enterprise Architecture and Sustainability teams to map your organization's existing data infrastructure against the study's two-layered framework, explicitly linking your micro-level AI model choices with macro-level organizational capabilities to support circular workflows.
  • IT Manager: Initiate a pilot project on your team's backlog to deploy a predictive maintenance model on a single class of high-wear manufacturing assets, using telemetry data to forecast asset degradation and automatically trigger life-extending maintenance tickets.
  • Business Strategist: Conduct a value-chain audit to identify where materials are lost, and draft a business case for integrating computer vision and robotic sorting systems at your primary waste-handling facility to automate material classification and close resource loops.
  • Researcher: Draft a research proposal for a 12-month longitudinal field study in partnership with a local electronics manufacturer to empirically validate the paper's four synthesized meta-principles in a real-world operational environment, addressing the literature's lack of longitudinal testing.
  • Policymaker: Draft a policy memo proposing a targeted grant program that provides R&D tax credits to industrial recycling facilities that implement integrated computer vision and automated sorting systems to improve regional material recycling rates.
Circular economy, Artificial intelligence, Design knowledge, Meta-Synthesis, Design science research, Dynamic capabilities
ORCHESTRATION PARADOXES IN NATIONAL QUANTUM COMPUTING INNOVATION ECOSYSTEMS
AI Processed Human Approved

What competing tensions and trade-offs do leaders face when guiding national quantum computing ecosystems?

Jori Taipale, Olli Tyrväinen, Tuure Tuunanen, Joel Mero, Teiko Heinosaari
ORCHESTRATION PARADOXES IN NATIONAL QUANTUM COMPUTING INNOVATION ECOSYSTEMS
ECIS 2026 · 2026
Key findings
  • Organizing tensions were identified as the most frequent paradox, specifically focusing on the trade-offs between collaboration vs. resource competition and guidance vs. flexibility.
  • Performance tensions emerged regarding resource allocation between software vs. hardware development and building a shared vision vs. individual firm survival.
  • Belonging tensions highlighted a 'godfather' paradox, where the ecosystem must balance the state's role as a major supportive funder and customer with the ultimate goal of corporate independence.
  • Learning tensions were characterized by the challenge of managing multi-talent knowledge development versus narrow technology specialization.
  • The study is limited by its single qualitative case context in Finland with a small sample of 15 informants, requiring caution when applying these findings to other national contexts.
What it means for you
  • CIO / IT Executive: Schedule a 9:00 AM meeting with your lead enterprise architect to audit your current quantum-readiness roadmap, establishing a formal policy that divides software development and hardware access investments 50/50 to hedge against the performance tension of committing prematurely to a single unproven hardware stack.
  • IT Manager: Create a bi-weekly 1-hour 'Quantum Cross-Training' slot in your team's calendar starting this Friday, pairing your specialized quantum physics PhDs with your generalist cloud engineers to address the learning tension between narrow technical specialization and broad-based systems integration.
  • Business Strategist: Perform a revenue dependency audit to calculate the percentage of your quantum business unit's funding derived from state grants versus commercial clients, and draft a pitch deck for two private-sector pilots to actively transition away from 'godfather' state-funding reliance.
  • Researcher: Draft a pre-competitive collaboration agreement template to send to a peer researcher at a competing institution, outlining a shared framework for baseline algorithm testing while clearly defining IP boundaries to manage the tension of collaboration versus resource competition.
  • Policymaker: Draft a revision for the upcoming national quantum funding round guidelines that introduces a 'co-investment requirement,' mandating that state-funded consortia secure at least 25% of their budget from private corporate buyers to counter the 'godfather' dependency paradox.
Ecosystem orchestration, Paradox theory, Quantum computing innovation ecosystem, Explorative case study, Interpretive qualitative research
TOWARDS DESIGN PRINCIPLES FOR TEACHING INDUSTRY 4.0 COMPETENCIES TO CS/IS STUDENTS
AI Processed

How can educators effectively teach modern industrial software skills without expensive physical equipment?

Felix Salcher
TOWARDS DESIGN PRINCIPLES FOR TEACHING INDUSTRY 4.0 COMPETENCIES TO CS/IS STUDENTS
ECIS 2026 · 2026
Key findings
  • Eight design principles were established, emphasizing virtualization, containerization, and problem-based learning cases as alternatives to expensive learning factories.\n- Preliminary course evaluation yielded positive descriptive feedback, with a mean rating of 4.43 out of 5 across satisfaction categories.\n- No formal statistical significance tests were performed due to the pilot nature of the study and the small, specific evaluation sample.\n- The pilot is limited by a very small sample size (9 respondents from a single cohort) and relies on subjective student perceptions rather than objective competency-based assessments.\n- Students noted difficulty adapting to the course's fast pace and programming requirements, highlighting a need for aligned prerequisites.
What it means for you
  • CIO / IT Executive: Direct your enterprise architecture team to draft a transition plan to replace physical IT hardware training labs with virtualized, containerized sandbox environments (e.g., Docker/Kubernetes) to reduce infrastructure costs and align onboarding with modern Industry 4.0 software architectures.
  • IT Manager: Review the programming prerequisites of your junior developers and assign them a structured, hands-on troubleshooting task using a containerized microservice mock-up on Monday morning to assess their readiness for complex Industry 4.0 integration projects.
  • Business Strategist: Reallocate a portion of next quarter's talent development budget away from high-cost physical equipment training and toward scalable, software-based virtual simulators to cost-effectively build digital twin and data architecture competencies.
  • Researcher: Draft a research proposal to conduct a multi-institution comparative study on these eight design principles, incorporating objective pre- and post-intervention technical testing rather than relying on subjective student feedback to measure competency gains.
  • Policymaker: Create a state-level educational grant initiative that incentivizes higher education institutions to update computer science curricula with virtualization-based Industry 4.0 frameworks, shifting funding priority away from physical learning factories.
Design Principles, Design Science Research, Industry 4.0, Smart Manufacturing
SUPPORT OR SUBSTITUTION: HOW CHATGPT AFFECTS COGNITIVE LOAD, OWNERSHIP, AND PERFORMANCE IN PROBLEM-SOLVING
AI Processed

How does using ChatGPT affect worker performance, workload, and their sense of ownership over problem-solving tasks?

Leonie Rebecca Freise, Josephine Mago Moritz
SUPPORT OR SUBSTITUTION: HOW CHATGPT AFFECTS COGNITIVE LOAD, OWNERSHIP, AND PERFORMANCE IN PROBLEM-SOLVING
ECIS 2026 · 2026
Key findings
  • Both ChatGPT-only access and ChatGPT with a prompting guide significantly improved task performance compared to the control group (Supported).\n- ChatGPT access significantly reduced temporal demand, but neither ChatGPT condition significantly reduced mental demand, and there was no significant effect of cognitive load on task performance (Partially Supported).\n- Both experimental conditions significantly reduced psychological ownership of the task outcome (Supported).\n- Reduced psychological ownership did not significantly impair task performance in this study (Not Supported).\n- Higher AI literacy significantly moderated and weakened the negative effects of AI access on temporal demand, and partially weakened the negative effect of the prompting guide on psychological ownership (Partially Supported).\n- The study's limitations include a modest sample size of 136 participants, a single-session experimental setting, and a reliance on self-reported measures.
What it means for you
  • CIO / IT Executive: Sign off on the procurement and scheduling of a mandatory 'AI Literacy and Collaborative Prompting' training program for all departments currently using generative AI, rather than just distributing licenses, to actively counter the erosion of psychological ownership and help staff manage cognitive load.
  • IT Manager: Schedule a 30-minute team meeting to establish a 'Human-in-the-Loop' workflow review protocol, requiring team members to explicitly document, edit, and present their personal contributions to any ChatGPT-generated deliverables to re-establish psychological ownership over their outcomes.
  • Business Strategist: Draft a proposal to revise departmental key performance indicators (KPIs), shifting the metrics of success from speed and volume (temporal savings) to qualitative, human-curated excellence to prevent employees from disengaging and treating their roles as outsourced, low-ownership tasks.
  • Researcher: Write a research proposal for a multi-week longitudinal field study with a cohort of at least 250 participants to track whether the observed short-term reduction in psychological ownership from AI usage leads to long-term employee disengagement or performance declines over time.
  • Policymaker: Publish a revised corporate governance policy that defines a 'Responsible Authorship' standard, mandating that the human employee remains 100% accountable for the accuracy and quality of AI-assisted outputs to prevent the dilution of organizational responsibility.
Generative Artificial Intelligence, Human-AI Collaboration, Cognitive Load, Psychological Ownership, AI Literacy, Task Performance
THE ANATOMY OF A DESIGN CHALLENGE FOR INITIATING AI INNOVATION
AI Processed Human Approved

How should teams structure initial challenges to successfully launch AI projects?

Franziska Anna Röckel, Benjamin van Giffen, Andreas Janson, Jennifer Hehn
THE ANATOMY OF A DESIGN CHALLENGE FOR INITIATING AI INNOVATION
ECIS 2026 · 2026
Key findings
  • The study empirically validated a structural anatomy for the Design Challenge consisting of four robust elements (verb, object of innovation, target group, context) and three plastic elements (internal data, technology, potential), which was supported across all 42 cases analyzed.
  • The robust elements were validated to successfully create a stable, shared identity and common direction among diverse team members, acting as a center of authority.
  • The plastic elements were shown to successfully absorb technological uncertainty, providing a flexible framework for teams to negotiate and adapt to AI's emergent capabilities over time.
  • The qualitative analysis identified three distinct nuances in shaping the 'object of innovation' (modular, abstract, and space) that guide teams toward either incremental or more radical innovation pathways.
What it means for you
  • CIO / IT Executive: Mandate that all upcoming AI project proposals use a standardized 'Design Challenge' charter template structured around the 4 robust elements (verb, object of innovation, target group, context) and 3 plastic elements (internal data, technology, potential), explicitly forbidding teams from locking in a rigid technical architecture in their initial pitch.
  • IT Manager: Gather your interdisciplinary project team for a 60-minute workshop on Monday morning to co-create a 'Design Challenge' statement, explicitly defining the stable 'robust' elements to align the business and tech members, while leaving the 'plastic' elements (like specific data sources and AI models) open as flexible targets to be negotiated as the project progresses.
  • Business Strategist: Review your company's AI project portfolio on Monday morning and intentionally rewrite the 'object of innovation' for each project: frame them as 'modular' if you need predictable, incremental improvements, or rewrite them as 'space-based' (e.g., focused on broad capabilities) if you want to push the team toward radical, disruptive business model innovation.
  • Researcher: Update your qualitative coding scheme for ongoing field studies of AI adoption to track how frequently and at what project stages teams modify the 'plastic' elements (internal data, technology, potential) versus the 'robust' elements of their project charters to measure the empirical threshold where technological uncertainty triggers project misalignment.
  • Policymaker: Revise the application guidelines for the upcoming public AI innovation grant cycle to replace rigid 'technical specifications' requirements with a mandatory 'Design Challenge' framework, scoring applicants on the clarity of their robust elements (target group, context) and the adaptive flexibility of their plastic elements (data and technology integration plans).
Design Challenge, AI Innovation, Boundary Object, Anatomy, Interdisciplinary Collaboration, Problem Formulation
Load More Showing 72 of 79