The concept of cognitive decline in AI systems has been a topic of debate, with some researchers suggesting that large language models (LLMs) and chatbots may experience a decline in cognitive abilities over time, similar to humans. However, others argue that this comparison is misleading due to anthropomorphization, and that AI models are frequently updated and improved, making it unfair to compare older and newer versions.
A recent study published in the BMJ found that LLMs, including ChatGPT and Gemini, performed poorly on visuospatial and executive function measures, with earlier versions scoring lower than more recent models. While these results may be concerning, it's essential to note that AI systems are not human brains and should not be diagnosed with cognitive decline or dementia.
The study's authors acknowledge that their findings challenge the assumption that AI will soon replace human doctors, particularly in fields that rely heavily on interpreting complex visual scenes. Instead, they suggest that AI systems should be viewed as tools that can assist and augment human capabilities, rather than replace them.
Ultimately, the debate surrounding AI's alleged cognitive decline highlights the need for ongoing research and development to improve the capabilities and reliability of AI systems.
