On humans and AI: A financial reporting dilemma

Bertomeu, Jeremy and Cheynel, Edwige and Lunawat, Radhika and Milone, Mario (2026): On humans and AI: A financial reporting dilemma.

Abstract

This study examines the resolution of ethical dilemmas in financial reporting by human participants and large language models. Participants act in the role of a CFO deciding whether to discontinue a prior policy with biased reporting; however, the bias is known and corrected by investors whereas a change may temporarily mislead investors. We find that models are less amenable to competing ethical considerations than humans, and exhibit greater preference for truthful reporting. Moreover, they respond with greater consistency to institutional ethical guidance, while humans become more indecisive under pressure from management. The models exhibit more internal coherence between their moral judgment and their policy prescriptions and are judged more persuasive by humans. Finally, humans follow model advice when accompanied by an explanation, but they seem to discount (and sometimes react against) advice offered without it. Our findings offer evidence on the misalignment between artificial intelligence and humans in tackling subjective reporting dilemmas while guiding the incorporation of such tools into corporate governance.

• Item Type: MPRA Paper
• Language: English
• Keywords: Artificial Intelligence, Ethics, Decision Making, Truth, Lies, Deception, Large Language Models, Financial Reporting, Experimental Accounting
• Subjects: C - Mathematical and Quantitative Methods > C9 - Design of Experiments > C91 - Laboratory, Individual Behavior; D - Microeconomics > D8 - Information, Knowledge, and Uncertainty > D83 - Search ; Learning ; Information and Knowledge ; Communication ; Belief ; Unawareness; M - Business Administration and Business Economics ; Marketing ; Accounting ; Personnel Economics > M4 - Accounting and Auditing > M41 - Accounting; M - Business Administration and Business Economics ; Marketing ; Accounting ; Personnel Economics > M4 - Accounting and Auditing > M48 - Government Policy and Regulation; O - Economic Development, Innovation, Technological Change, and Growth > O3 - Innovation ; Research and Development ; Technological Change ; Intellectual Property Rights > O33 - Technological Change: Choices and Consequences ; Diffusion Processes

References

Bibliography

Achiam et al. (2024): GPT-4 Technical Report. arXiv:2303.08774. Aher, Arriaga, Kalai (2023): Using LLMs to Simulate Multiple Humans. ICML. Aobdia, Ma, Zhang (2025): Auditing Effects on Employment Hiring. SSRN 5703842. Argyle et al. (2023): Out of One, Many: LLMs to Simulate Human Samples. Political Analysis 31(3). Awad et al. (2018): The Moral Machine experiment. Nature 563. Balesni et al. (2024): Towards evaluations-based safety cases for AI scheming. arXiv:2411.03336. Basu (1997): The conservatism principle. JAE 24(1). Bentley et al. (2021): Identifying Insincere and Sincere Bias. TAR 96(5). Bertomeu and Cheynel (2026): Truth and Deception. SSRN 6471958. Bloomfield and Rennekamp (2009): Experimental Research in Financial Reporting. F&T Accounting 3(1). Bloomfield (2021): Moral Accountability Principles. SSRN 3812505. Brand, Israeli, Ngwe (2023): Using LLMs for Market Research. SSRN 4395751. Cardinaels and Yin (2015): Think Twice Before Going for Incentives. JAR 53(5). Chang et al. (2026): AI Democratization and Trading Inequality. JAR (Forthcoming). Chen et al. (2025): A Manager and an AI Walk into a Bar. M&SOM 27(2). Chen et al. (2023): The emergence of economic rationality of GPT. PNAS 120(51). Choi and Xie (2025): Human + AI in Accounting. SSRN 5240924. Davidson and Stevens (2013): Can a Code of Ethics Improve Manager Behavior? TAR 88(1). Dickhaut et al. (2010): Neuroaccounting. Accounting Horizons 24(2). Dietvorst, Simmons, Massey (2015): Algorithm aversion. J Exp Psychol 144(1). Dye (1988): Earnings Management in an Overlapping Generations Model. JAR 26(2). Emett et al. (2025): Leveraging ChatGPT for Internal Audit. Accounting Horizons 39(2). Eulerich et al. (2024): Is it all hype? ChatGPT in Accounting. RAST 29(3). Evans et al. (2001): Honesty in Managerial Reporting. TAR 76(4). Fischer and Huddart (2008): Optimal Contracting with Endogenous Social Norms. AER 98(4). Friedman (2014): Implications of power: CEO pressure on CFO. JAE 58(1). Gazal Ayal, Elyoseph, Solomon (2026): Evaluating LLMs as Judicial Decision-Makers. Justice Quarterly. Gibson, Tanner, Wagner (2013): Preferences for Truthfulness. AER 103(1). Gneezy, Rockenbach, Serra-Garcia (2013): Measuring lying aversion. JEBO 93. Greenblatt et al. (2024): Alignment faking in LLMs. arXiv:2412.14093. Hennes, Leone, Miller (2008): Distinguishing Errors from Irregularities. TAR 83(6). Horton, Filippas, Manning (2026): LLMs as Simulated Economic Agents. arXiv:2301.07543. Hubinger et al. (2024): Sleeper Agents. arXiv:2401.05566. Huddart and Qu (2024): Rotten Apples and Sterling Examples. JMAR 36(2). Indjejikian and Matejka (2009): CFO Fiduciary Responsibilities. JAR 47(4). Jarviniemi and Hubinger (2024): Uncovering Deceptive Tendencies in LMs. arXiv:2405.01576. Kahneman and Tversky (1979): Prospect Theory. Econometrica 47(2). Karatas and Cutright (2023): Thinking about God and AI acceptance. PNAS 120(33). Kartik, Ottaviani, Squintani (2007): Credulity, lies, and costly talk. JET 134(1). Kim, Muhn, Nikolaev (2025): Bloated Disclosures. arXiv:2306.10224. Kim, Liang, Hooker (2024): Yuji Ijiri's Fairness Question. Accounting, Economics, and Law. Kirshner (2024): Artificial Agents in Operations Management Experiments. SSRN 4726933. Kohlberg (1981): The Philosophy of Moral Development. Harper & Row. Leng (2024): Folk Economics in the Machine. SSRN 4705130. Levy (2024): Caution Ahead: Numerical Reasoning and Look-ahead Bias. SSRN 5082861. Li et al. (2025): The Promise and Peril of Generative AI. arXiv:2412.01069. Li, Castelo, Katona, Sarvary (2024): Validity of LLMs for Perceptual Analysis. Marketing Science 43(2). Loewenstein and Wojtowicz (2025): The Economics of Attention. JEL 63(3). Logg, Minson, Moore (2019): Algorithm appreciation. OBHDP 151. Ma, Zhang, Saunders (2023): Is ChatGPT Humanly Irrational? Macmillan-Scott and Musolesi (2024): (Ir)rationality and Cognitive Biases in LLMs. arXiv:2402.09193. Mei et al. (2024): A Turing test of whether AI chatbots are behaviorally similar to humans. PNAS 121(9). Meinke et al. (2025): Frontier Models are Capable of In-context Scheming. arXiv:2412.04984. Ottaviani and Squintani (2006): Naive audience and communication bias. IJGT 35(1). Ouyang et al. (2022): Training LMs to follow instructions with human feedback. NeurIPS. Posner and Saran (2025): Judge AI: Assessing LLMs in Judicial Decision-Making. SSRN 5098708. Scheurer, Balesni, Hobbhahn (2024): LLMs can Strategically Deceive their Users. arXiv:2311.07590. Simon (2019): The Sciences of the Artificial. MIT Press. Sobel (2020): Lying and Deception in Games. JPE 128(3). Stein (1989): Efficient Capital Markets, Inefficient Firms. QJE 104(4). Sunder (2005): Minding our manners: Accounting as social norms. British Accounting Review 37(4). Suri, Slater, Ziaee, Nguyen (2023): Do LLMs Show Decision Heuristics Similar to Humans? arXiv:2305.04400. Tayler and Bloomfield (2011): Norms, Conformity, and Controls. JAR 49(3). Vasarhelyi et al. (2023): LLMs in Accounting. JETA 20(2). Watts (2003): Conservatism in Accounting Part I. Accounting Horizons 17(3). Wood et al. (2023): The ChatGPT Chatbot: Accounting Assessment Questions. IAE 38(4). Zhang and Gosline (2023): Human favoritism, not AI aversion. JDM 18.