This article explores how generative artificial intelligence programs can write convincingly yet struggle to evaluate basic scientific statements reliably. Testing across multiple prompts and versions shows inconsistent and biased performance, raising questions about use in decision making and the need for human oversight.
The piece highlights why attention to reliability and bias in AI is of interest to psychology by illustrating how human judgment remains essential when evaluating reasoning and evidence, and by showing how pattern recognition and language fluency can mask underlying cognitive limits in automated systems.
Article Title: Artificial intelligence struggles to consistently evaluate scientific facts
Link to PsyPost Article: https://www.psypost dot org/artificial-intelligence-struggles-to-consistently-evaluate-scientific-facts/
https://www.psypost dot org/artificial-intelligence-struggles-to-consistently-evaluate-scientific-facts/
Copy and paste broken link above into your browser and replace "dot" with "." for link to work. We have to do it this way to avoid displaying copyrighted images.
#AI #cognition #bias #evaluation #psychology