A recent study raises concerns that older AI models, like humans, can experience cognitive decline with age. Researchers tested several large language models, including OpenAI's ChatGPT, Anthropic's Sonnet, and Alphabet's Gemini, using the Montreal Cognitive Assessment (MoCA) test, which is commonly used to assess cognitive impairment in humans.
The results showed that while the latest version of ChatGPT scored well, older models performed poorly, particularly in visuospatial and executive function tasks. For example, Gemini 1.0 scored only 16 out of 30, indicating significant cognitive impairment.
These findings have significant implications for the use of AI in medical diagnostics and other applications where reliability and accuracy are crucial. The study's authors suggest that AI models may need to be regularly updated and evaluated to ensure they remain effective and reliable.
The study also highlights the need for further research into the cognitive limitations of AI models and how they can be addressed. As AI becomes increasingly integrated into various aspects of our lives, understanding its potential vulnerabilities is essential.
