Can an AI Make You Doubt Your Own Judgement?
Disagreement from a machine can linger even after we reject its advice.
Author
Alok Jha
Reading
2 min read
Category
AI + Human
Suppose you examine a photograph and decide it is authentic. An AI tool says it is manipulated. You look again and still disagree. Has nothing changed? Maybe not. You may leave the interaction slightly less confident than you entered it.
Emerging research suggests AI advice can influence not only decisions but confidence in human judgement. This matters because confidence affects whether we seek second opinions, defer decisions or learn from feedback.
Machines have a peculiar authority. Their outputs often appear numerical, technical or data-driven. Even when we know systems make mistakes, disagreement can produce a small internal question: 'What did it see that I missed?'
That question is healthy when AI has genuine information advantage. It becomes problematic when people systematically discount their own expertise merely because the opposing view came from a machine.
The reverse problem also exists: algorithm aversion. One visible AI mistake can make users reject useful systems altogether. Good human-AI judgement requires avoiding both worship and sulking.
A practical technique is to record your initial judgement and confidence before viewing AI advice in high-stakes tasks. After seeing the recommendation, note what specific evidence changed your mind. This separates update from intimidation.
Organisations can design for this by asking employees to make independent assessments before AI output is revealed. Otherwise the first displayed recommendation becomes an anchor.
AI should make us better calibrated, not merely more obedient or more defensive. The important question after disagreement is not 'Who won?' but 'What evidence deserves to change confidence, and by how much?'
Selected research anchors & further reading
Pearson, J. et al. (2026). Examining human reliance on artificial intelligence in decision making. Scientific Reports. Source
Schilke, O., & Reimann, M. (2025). The transparency dilemma: How AI disclosure erodes trust. OBHDP. Source
Phan, K. D., & Truong-Dinh, B. Q. (2026). When conversational AI personalises too much. Computers in Human Behavior. Source
Lee, R. S., De Silva Kanakaratne, M., & van der Veen, R. (2026). Consent management and chatbot anthropomorphism. Journal of Business Research. Source
AI or human: How the type of information to be disclosed alters customer service agent preferences (2026). Source
About Human Signals
Human Signals is Alok Jha's personal publication about psychology, behaviour and the choices we make. It explores the mind, decision-making, money, business and the changing relationship between humans and AI.
The editorial ambition is simple: explain people without preaching to them. Evidence matters, but so do everyday details—the delayed phone call, the abandoned shopping cart, the retirement account someone is afraid to spend, the employee who stays silent in a meeting, the AI answer that sounds more certain than it should.
About Alok Jha
Alok Jha is an entrepreneur, business strategist, mentor and student of human behaviour. He has more than three decades of corporate and CXO-level experience across fintech, technology, startups and new business development, along with postgraduate study in management and psychology. Human Signals brings together those two long-running interests: how people behave, and what that behaviour changes in real life and business.