Ethan Mollick is an American academic, professor at the Wharton School of the University of Pennsylvania, and widely cited public intellectual on the practical deployment of generative artificial intelligence in work, education, and entrepreneurship. His research bridges management science and AI-adoption theory, investigating how large language models augment human productivity, reshape organisational workflows, and alter pedagogical practice. He is the author of the book ‘Co-Intelligence: Living and Working with AI’ and is known for his hands-on experimental approach to understanding AI capabilities and their societal implications.
Semantic Classification
Content
Compositional Relationships (Components)
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:hasPart ai:CoIntelligenceFramework))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:hasPart ai:JaggedTechnologicalFrontier))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:hasPart ai:OneUsefulThingNewsletter))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:hasPart ai:WhartonGenerativeAILab))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:hasPart ai:EmpiricalAIResearch))
Dependency Relationships
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:requires ai:LargeLanguageModels))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:requires ai:GenerativeAI))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:dependsOn ai:WhartonSchool))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:dependsOn ai:FieldExperimentation))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:uses ai:PromptEngineering))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:uses ai:RandomisedControlledTrials))
Capability Relationships
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:enables ai:AIAdoption))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:enables ai:AILiteracy))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:enables ai:HumanAICollaboration))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:enables ai:KnowledgeWorkAugmentation))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:enables ai:AIInEducation))
Implementation Relationships
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:implements ai:CoIntelligenceFramework))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:supports ai:Entrepreneurship))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:supports ai:InnovationManagement))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:supports ai:OrganisationalLearning))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:supports ai:FutureOfWork))
Reduction Relationships
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:reducesTo ai:ManagementScience))
SubClassOf(ai:EthanMollick
ObjectSomeValuesFrom(ai:reducesTo ai:AIAdoptionTheory))
About
Ethan Mollick’s career trajectory illustrates the convergence of management science, entrepreneurship research, and artificial intelligence that characterises a distinctive strand of applied AI scholarship in the 2020s. He completed his PhD at MIT Sloan where he worked on the economics of innovation and the structure of firms — research that resulted in papers on the role of middle management in innovation, the emergence of crowdfunding as a financing mechanism, and the distribution of talent in creative industries. These early contributions, focused on how organisations produce and deploy creative and innovative work, laid the intellectual groundwork for his later research on AI as a transformative technology affecting the same processes.
His turn toward AI research accelerated dramatically with the release of ChatGPT in late 2022. Where many academic economists took a cautious wait-and-see posture, Mollick engaged immediately and publicly with large language model capabilities, conducting structured informal experiments and publishing results in real time through his “One Useful Thing” Substack. This newsletter grew to over 200,000 subscribers by 2024 and became a major vehicle for translating AI research findings into actionable guidance for professionals, educators, and managers. His approach combined personal experimentation — using AI tools extensively in his own teaching, writing, and research — with formal research designs, producing a distinctive hybrid of practitioner-oriented commentary and rigorous empirical work that distinguished him from both pure theorists and popular technologists.
The most significant empirical contribution of this period was the “Navigating the Jagged Technological Frontier” study (2023, published in Organization Science 2025), conducted with Fabrizio Dell’Acqua, Edward McFowland III, and others, in collaboration with Boston Consulting Group. The study randomised approximately 758 BCG consultants to either use ChatGPT-4 or not, measuring performance across 18 diverse tasks replicating real consulting work. The results showed that AI access raised average performance by 12–18% across tasks within the frontier of AI capability, but degraded performance for tasks outside that frontier — a result with significant implications for how organisations should think about AI deployment, task allocation, and human skill development.
Mollick co-directs the Wharton Generative AI Labs (GAIL) with Lilach Mollick, a Senior Research Fellow focused on pedagogical applications of AI. GAIL builds prototypes and conducts research on how AI can help humans thrive while mitigating risks, with a particular emphasis on educational applications and the development of AI tutors that combine the accessibility of conversational AI with evidence-based pedagogical principles. By 2026 Mollick had extended his research focus to agentic AI systems — AI agents that can autonomously complete multi-step tasks over hours or days — presenting findings at the 2026 AI and the Workforce Summit showing that small improvements in model accuracy multiply agentic task capacity by 2–3x, with fundamental implications for the allocation of knowledge work between humans and AI systems.
Key Frameworks and Contributions
The Jagged Technological Frontier Mollick’s central theoretical contribution. AI capabilities are not uniformly distributed across task types. Within the frontier — tasks where AI is capable — performance is dramatically augmented. Outside the frontier — tasks where AI fails in ways that are non-obvious — AI involvement degrades outcomes. The boundary between these is “jagged”: counterintuitive, rapidly shifting, and requiring empirical investigation rather than assumption. Implications include the danger of over-reliance on AI for tasks it handles poorly and the risk of under-utilisation for tasks it handles well. The concept has been widely adopted in management consulting, HR strategy, and AI policy discourse.
The Co-Intelligence Framework Developed in his 2024 book, the co-intelligence framework proposes that AI should be understood as a new form of intelligence with whom humans co-labour, rather than as a tool, an oracle, or a replacement. This involves treating AI as always on (constantly available), as a brilliant friend (rather than a formal service provider), and as requiring users to maintain critical oversight. The framework addresses four key challenges: how to use AI, how to preserve human judgment, how to avoid cognitive surrender, and how to maintain authentic creativity and learning while using AI assistance.
Cognitive Surrender A concept developed with Wharton colleagues describing the tendency of humans working with AI to stop thinking critically and delegate decision-making entirely to AI outputs, even when those outputs are incorrect. Mollick’s research showed that users who asked AI to explain its reasoning, or used AI for only part of a task, were significantly less susceptible to this failure mode. The concept has parallels to automation bias in human-factors psychology and has become central to discussions of responsible AI adoption in high-stakes domains.
AI as a General-Purpose Technology Mollick’s position — shared with economists Erik Brynjolfsson and Daron Acemoglu but with a distinctive emphasis on near-term practical management — is that generative AI is a general-purpose technology comparable to electrification or the internet, implying a productivity J-curve: initial disruption before complementary innovations and practices produce widespread productivity gains. His empirical work showing heterogeneous effects across tasks and workers grounds this theoretical claim in field data.
Experimental Pedagogy for AI Literacy With Lilach Mollick, he developed AI tutoring systems and pedagogical frameworks for integrating AI into university education in ways that promote rather than undermine learning. This includes “mentor mode” prompting strategies in which AI is instructed to ask questions rather than give answers, and structured reflective exercises that prevent students from bypassing learning by outsourcing thinking to AI systems.
Applications and Use Cases
Professional Upskilling and Workforce Development Mollick’s writing directly informs how individual professionals approach AI Adoption — providing worked examples of AI-assisted task completion, comparisons between AI-naive and AI-augmented performance, and tactical guidance on Prompt Engineering and task decomposition. His influence on management consulting, legal, financial, and creative knowledge workers is substantial.
Educational Policy and Curriculum Reform His research and public advocacy have shaped debate on AI in universities — how to redesign assessments in the age of AI, how to use AI as a pedagogical tool rather than a cheating shortcut, and how to develop AI literacy as a transferable professional skill. His findings on AI’s harmful effects on learning when used without reflection (referenced in PNAS 2025 studies on generative AI in mathematics education) inform debates on academic integrity.
Organisational AI Strategy The BCG study and subsequent work have influenced how large organisations think about AI deployment strategy — which tasks to automate, which to augment, and which to preserve as human-led. Mollick’s concept of the jagged frontier has become a practical framework in management consulting and HR strategy discussions about AI adoption.
Management Practice and the Manager as AI Superpower Mollick has argued that effective managers who understand how to deploy AI agents — directing, monitoring, correcting — will gain an asymmetric advantage, as management skills (goal-setting, quality assessment, delegation) transfer directly to AI orchestration. His 2026 research on agentic AI systems (“Real AI Agents and Real Work,” One Useful Thing) explored this theme directly.
AI Policy and Public Discourse As a credible academic voice who engages empirically rather than speculatively, Mollick’s work informs AI Policy debates about workforce readiness, education reform, and the governance of AI deployment in knowledge-intensive sectors. He has been cited in US Congressional briefings, UK government AI strategy documents, and OECD reports on AI and the future of work.
Entrepreneurship and the Lowered Barrier to Innovation His background in Entrepreneurship research converges with his AI work: generative AI lowers the cost of key entrepreneurial tasks (market research, prototyping, copywriting, business planning), potentially democratising venture creation. His research explores both the positive and negative implications of this for innovation ecosystems.
Startups and AI-Native Product Development Practitioners in AI-native startups draw on Mollick’s frameworks for thinking about when AI augments versus replaces, and how to design human-AI workflows that preserve the complementarity between human judgment and machine throughput.
Academic Context
Mollick holds a tenured position in Wharton’s Management Department. His early academic work covered crowdfunding (including some of the first rigorous empirical studies of Kickstarter dynamics), the role of middle management in large firm innovation, and the distribution of creative talent in the games industry. He published in top management journals including Management Science, Organization Science, and Strategic Management Journal before pivoting to AI research.
The AI productivity research he has contributed to sits within a broader empirical literature including Brynjolfsson, Li and Raymond’s “Generative AI at Work” (studying GitHub Copilot at a customer-service firm, QJE 2025), Dell’Acqua et al. on the BCG study, and Noy and Zhang’s study of ChatGPT for writing tasks. Mollick’s distinctive contribution has been large-scale randomised field experiments in professional settings with diverse task batteries, rather than narrow domain studies.
His work intersects with foundational theories of technology and productivity from economics (Brynjolfsson’s productivity paradox, Acemoglu-Restrepo’s task-based models of automation) and with management theories of organisational learning and knowledge management. The co-intelligence framework draws philosophically on Douglas Engelbart’s augmentation thesis (1962), which argued for using computers to augment human intellect rather than replace it, and on distributed cognition theory from Cognitive Science.
Mollick’s influence on the AI discourse in academia and practice has been amplified by his position at Wharton — one of the world’s most influential business schools — and by TIME Magazine naming him among the 100 Most Influential People in AI in 2024 (alongside Demis Hassabis, Sam Altman, and Dario Amodei).
Current Landscape (2026)
By mid-2026 Mollick’s research focus has shifted substantially toward agentic AI — systems capable of completing extended multi-step tasks with minimal human intervention. At the 2026 AI and the Workforce Summit he presented research showing that even small accuracy improvements in underlying models translate to exponential expansion in the number of tasks an AI agent can complete without human error-correction, because error accumulation across long task chains is highly sensitive to per-step error rates. This has profound implications for how organisations structure knowledge work.
Wharton Generative AI Labs tracking studies report 75% of companies identifying positive ROI from AI initiatives by early 2026, a significant increase from earlier adoption cycles. Mollick has characterised this as the beginning of the productivity J-curve recovery, with complementary innovations in workflow design, AI tooling, and human-AI interface maturation beginning to yield aggregate gains.
His Substack “One Useful Thing” published influential pieces in 2025–26 including “Thinking Like an AI” (exploring how understanding AI’s token-prediction mechanism helps users work with rather than against its failure modes), “Real AI Agents and Real Work” (documenting practical agentic AI deployment), “Management as AI Superpower” (arguing for managerial skills as transferable to AI orchestration), and “Choosing to Stay Human” (addressing cognitive surrender and the ethics of AI augmentation in creative work). These reached audiences of hundreds of thousands and were widely cited in corporate AI strategy discussions.
His CNBC interview (October 2025) attracted attention for his frank statement — “No one knows anything” — regarding AI job displacement predictions, positioning him as a credible voice of empirical humility in a discourse dominated by confident but poorly-grounded forecasts.
UK Context
While Ethan Mollick is an American academic based at Wharton, his work has had significant uptake in the UK AI ecosystem. His “jagged frontier” framework has been referenced in UK government AI strategy discussions, including inputs to the Department for Science, Innovation and Technology (DSIT) and the Cabinet Office’s AI Opportunities Action Plan (2025). The co-intelligence framing has influenced UK university responses to AI in assessment, with the Quality Assurance Agency (QAA) citing research in his tradition when updating academic integrity guidance.
UK business schools — particularly London Business School, Said Business School (Oxford), Judge Business School (Cambridge), and Warwick Business School — have engaged with Mollick’s empirical productivity research in their AI strategy and management education curricula. His BCG study has been widely taught in UK MBA programmes as a case study in responsible AI deployment design.
The Alan Turing Institute, which coordinates UK AI research strategy, has engaged with the Mollick tradition of field experimentation for AI productivity research. UK think tanks including the Tony Blair Institute for Global Change, IPPR, and Resolution Foundation have cited his work in labour market analyses of AI impact. His influence on the UK AI Skills Framework (published 2025) is notable through the concept of practical AI literacy as a transferable professional competency rather than a purely technical qualification.
Northern England context: in Manchester, Leeds, Sheffield, and Newcastle, where digital skills gaps and AI adoption in manufacturing and service sectors are policy priorities, Mollick’s accessible empirical framing of AI productivity has been adopted by organisations such as the Northern Powerhouse Partnership and Greater Manchester Combined Authority in their AI workforce readiness programmes. The emphasis on experimentation — trying AI tools directly rather than waiting for top-down guidance — resonates with the pragmatic innovation culture of Northern English SMEs.
Future Directions (2026–2030)
Agentic AI and the Transformation of Knowledge Work Mollick’s current research trajectory focuses on how agentic AI — AI that can execute long multi-step tasks, use tools, and self-correct — transforms job roles, organisational structure, and the economics of professional services. The shift from AI as assistant to AI as agent may require new management frameworks beyond the co-intelligence model.
AI Tutoring and Personalised Education at Scale GAIL is building and evaluating AI tutoring systems designed to deliver one-on-one Socratic dialogue at scale, potentially addressing the productivity paradox in education. Longitudinal studies of AI tutoring effects on learning and retention are in progress.
Measuring Cognitive Surrender at Population Scale Research on the conditions under which cognitive surrender occurs — and how to design AI interfaces that prevent it — is likely to become increasingly central as AI adoption deepens. Mollick and colleagues are developing measurement frameworks for assessing the quality of human-AI co-cognition rather than merely task output quality.
AI and Entrepreneurial Dynamism As generative AI lowers barriers to entry for many knowledge-intensive businesses, Mollick’s entrepreneurship background positions him to study the downstream effects on industry concentration, firm formation rates, and the distribution of innovation rents.
Governance and AI Policy Engagement The shift from AI as curiosity to AI as economic infrastructure will demand more formal engagement with governance questions. Mollick’s work may increasingly intersect with debates about AI regulation, liability, and the design of national AI skills strategies — areas where his empirical grounding provides a complement to the more speculative work dominating policy circles.
Research & Literature
- Mollick, E. (2024). Co-Intelligence: Living and Working with AI. Portfolio/Penguin Random House. ISBN 9780593716717.
- Dell’Acqua, F., McFowland, E., Mollick, E. et al. (2023 / published 2025). “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality.” Organization Science. https://pubsonline.informs.org/doi/10.1287/orsc.2025.21838
- Brynjolfsson, E., Li, D. and Raymond, L.R. (2023 / published 2025). “Generative AI at Work.” Quarterly Journal of Economics, 140(2), 889–. https://academic.oup.com/qje/article/140/2/889/7990658
- Noy, S. and Zhang, W. (2023). “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence.” Science, 381(6654), 187–192.
- Mollick, E. and Mollick, L. (2023). “Using AI to Implement Effective Teaching Strategies in Classrooms: Five Strategies, Including Prompts.” Wharton Working Paper.
- Mollick, E. (2023). “Why All Our Frameworks for AI Are (Probably) Wrong.” One Useful Thing, Substack. https://www.oneusefulthing.org/
- Mollick, E. (2025). “Real AI Agents and Real Work.” One Useful Thing, Substack. https://www.oneusefulthing.org/p/real-ai-agents-and-real-work
- Mollick, E. (2025). “Thinking Like an AI.” One Useful Thing, Substack. https://www.oneusefulthing.org/p/thinking-like-an-ai
- Mollick, E. (2025). “Management as AI Superpower.” One Useful Thing, Substack. https://www.oneusefulthing.org/p/management-as-ai-superpower
- Mollick, E. (2025). “Choosing to Stay Human.” One Useful Thing, Substack. https://www.oneusefulthing.org/p/choosing-to-stay-human
- Mollick, E. and Mollick, L. (2024). “Instructors as Innovators: A Future-Focused Approach to New AI Learning Opportunities, With Prompts.” PLOS ONE, 19(5), e0300024.
- Belwalkar, B. (2026). Review of Co-Intelligence. Personnel Psychology. https://onlinelibrary.wiley.com/doi/10.1111/peps.70012
- Shen, S. (2025). Review of Co-Intelligence. Administrative Science Quarterly. https://journals.sagepub.com/doi/10.1177/00018392251341268
- Mollick, E.R. (2012). “People and Process, Suits and Innovators: The Role of Individuals in Firm Performance.” Strategic Management Journal, 33(9), 1001–1015.
- Mollick, E.R. and Robb, A. (2016). “Democratizing Innovation and Capital Access: The Role of Crowdfunding.” California Management Review, 58(2), 72–87.
- Mollick, E.R. (2014). “The Dynamics of Crowdfunding: An Exploratory Study.” Journal of Business Venturing, 29(1), 1–16.
- Wharton Generative AI Labs. (2026). About Us. https://gail.wharton.upenn.edu/about-us/
- TIME Magazine (2024). “TIME100 AI: The 100 Most Influential People in AI.” Time Inc.
- Acemoglu, D. and Restrepo, P. (2022). “Tasks, Automation, and the Rise in US Wage Inequality.” Econometrica, 90(5), 1973–2016.
- Brynjolfsson, E., Rock, D. and Syverson, C. (2021). “The Productivity J-Curve: How Intangibles Complement General Purpose Technologies.” American Economic Journal: Macroeconomics, 13(1), 333–372.
- Valence Research. (2026). “AI Agents, Agentic Work and the Future of Work: Ethan Mollick.” https://www.valence.co/ai-and-the-workforce/ai-agents-agentic-work-the-future-of-work-ethan-mollick
- Engelbart, D. (1962). “Augmenting Human Intellect: A Conceptual Framework.” SRI Summary Report AFOSR-3223, Stanford Research Institute.
- Wharton Management Department. (2026). Ethan Mollick Faculty Profile. https://mgmt.wharton.upenn.edu/profile/emollick/
- Demirer, M. et al. (2024). “The Effects of Generative AI on High-Skilled Work.” MIT Working Paper. https://economics.mit.edu/sites/default/files/inline-files/draft_copilot_experiments.pdf
- Coachingforleaders.com. (2025). “Principles for Using AI at Work, with Ethan Mollick.” Podcast episode 674. https://coachingforleaders.com/podcast/principles-for-using-ai-ethan-mollick/
- Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics. (2025). PNAS, 122. https://www.pnas.org/doi/10.1073/pnas.2422633122
- Insight Partners. (2024). “The Jagged Frontier of Generative AI: A Conversation with Ethan Mollick.” https://www.insightpartners.com/ideas/generative-ai-ethan-mollick/
- CNBC. (2025). “Don’t Trust AI Jobs Predictions, Says Wharton Expert Ethan Mollick.” https://www.cnbc.com/2025/10/07/generative-ai-knowledge-learning-jobs-education-skills-wharton-ethan-mollick.html