Quick summary: Recent research into large language models reveals that while artificial intelligence can simulate human trust by evaluating competence and integrity, it does so through a rigid and fragmented scoring system that lacks human intuition. This systematic approach frequently amplifies demographic biases regarding age, gender, and religion, leading to significant disparities in critical financial and social decisions such as lending or recruitment. As these systems transition from assistants to primary decision makers, public policy and healthcare practice must implement rigorous oversight to ensure that machine trust does not move towards a future of predictable and automated discrimination.
Artificial intelligence systems do not just process information. They systematically judge people in ways that resemble human trust, but with important differences, according to a new study by researchers at the Hebrew University of Jerusalem.
The research, published in Proceedings of the Royal Society A by Professor Yaniv Dover and Valeria Lerman of the Hebrew University Business School, draws on more than 43,000 simulated decisions alongside nearly 1,000 human participants. It examines how today’s most advanced AI systems, including models similar to ChatGPT and Google’s Gemini, make judgements about people and form something that looks a lot like trust.
To test this, the study placed both humans and AI in familiar scenarios: deciding how much money to lend a small business owner, whether to trust a babysitter, how to rate a boss, or how much to donate to a nonprofit founder. A clear pattern emerged. Both humans and AI favoured people who seemed competent, honest, and well intentioned.
“AI is not making random decisions. It captures something real about how humans evaluate one another,” says Professor Dover.
The resemblance, however, stops there. Humans tend to blend multiple traits into a single, intuitive, holistic judgement. AI breaks people down into components, scoring competence, integrity, and kindness almost like separate columns in a spreadsheet, producing a more rigid, consistent, but less human style of evaluation.
“People in our study are messy and holistic in how they judge others,” explains Lerman. “AI is cleaner, more systematic, and that can lead to very different outcomes.”
A troubling pattern of amplified bias also emerged. In financial scenarios such as lending and donation decisions, AI systems showed consistent and sometimes sizeable differences based solely on demographic traits, even when every other detail about the person was identical. Key findings include:
- Age frequently influenced outcomes, with older individuals often receiving more favourable results, though the opposite pattern also appeared in some cases.
- Religion had a significant effect, particularly in monetary decisions across various models.
- Gender influenced decisions in certain models and scenarios.
“Humans have biases, of course. But what surprised us is that AI’s biases can be more systematic, more predictable, and sometimes stronger,” says Professor Dover.
The researchers also found that different AI models often made different judgements about the same person, with some systems rewarding a trait that others penalised.
“Which model you use really matters. Two systems can look similar on the surface but behave very differently when making decisions about people,” Lerman notes.
AI is already being used to screen job candidates, assess creditworthiness, recommend medical actions, and guide organisational decisions. As these systems move from assistants to decision makers, the researchers emphasise that their findings are not a warning against AI, but a call for awareness.
“These systems are powerful. They can model aspects of human reasoning in a consistent way. But they are not human, and we should not assume they see people the way we do,” says Professor Dover.
As AI becomes more embedded in everyday life, the question is no longer whether we trust machines. It is whether we understand how they trust us.
The paper, titled “A Closer Look at How Large Language Models ‘Trust’ Humans: Patterns and Biases,” was conducted as part of the new Research Centre for AI in Organisations at the Hebrew University School of Business.