What is it about?

Researchers tested four popular AI chatbots (ChatGPT, Claude, Gemini, and Grok) using personality tests normally given to humans. They found each AI has distinct personality traits - Claude acts more introverted and analytical, while ChatGPT is more extroverted and assertive. The study tested each AI 15 times to make sure the results were consistent. This matters because these AIs are increasingly being used in healthcare and mental health support, where personality can affect how well patients connect with and trust their digital helpers.

Featured Image

Why is it important?

This is one of the first studies to scientifically measure AI personality traits using validated psychology tests, which is important as healthcare systems rapidly adopt AI chatbots for patient support and mental health services. The research shows that AI isn't neutral - each system has built-in personality characteristics that could influence patient interactions, therapeutic relationships, and treatment outcomes. This challenges the assumption that AI tools are interchangeable and suggests that healthcare providers need to consider personality matching between AI systems and patients, much like the importance of therapist-patient compatibility in achieving successful treatment outcomes.

Perspectives

I find this research fascinating because it reveals something we've intuitively sensed but never formally proven - that different AI systems feel different to interact with. The consistency of Claude's INTJ profile across all tests is particularly striking, suggesting that these aren't random variations but rather stable characteristics. What concerns me is that healthcare organizations might deploy AI without considering these personality factors, potentially creating mismatches that could harm therapeutic relationships. I think this work raises entirely new questions about AI design ethics and whether we should deliberately engineer specific personalities for different medical contexts.

Thomas F Heston MD
University of Washington

Read the Original

This page is a summary of: Large Language Models Demonstrate Distinct Personality Profiles, Cureus, May 2025, Springer Science + Business Media,
DOI: 10.7759/cureus.84706.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page