Story perspectives

Study Reveals ChatGPT's 60% Accuracy in Science Claims

3/24/2026

Summary

Story summary

ChatGPT shows about 60% accuracy in evaluating scientific claims, according to a Rutgers Business Review study.
Researchers tested more than 700 claims and found ChatGPT's responses often oscillated between true and false.
Dr. Mesut Cicek, the lead researcher, emphasizes skepticism when using AI for nuanced tasks.
Cicek cautions that while AI can generate convincing language, it lacks true understanding and users should remain cautious of AI-generated information.

1 of 1

Story summary

ChatGPT shows about 60% accuracy in evaluating scientific claims, according to a Rutgers Business Review study.
Researchers tested more than 700 claims and found ChatGPT's responses often oscillated between true and false.
Dr. Mesut Cicek, the lead researcher, emphasizes skepticism when using AI for nuanced tasks.
Cicek cautions that while AI can generate convincing language, it lacks true understanding and users should remain cautious of AI-generated information.