Leading AI models excel at basic tasks but lack scientific reasoning: Study

A study published in Nature Computational Science reveals that while leading AI models excel in basic scientific tasks, they lack essential scientific reasoning capabilities, posing risks for research applications.

• The study, conducted by NM Anoop Krishn and his team from IIT Delhi and Florida State University, assesses the performance of leading AI models in scientific tasks.
• Published in the prestigious journal Nature, the research indicates that despite advancements, AI models fail in complex reasoning, which could hinder scientific progress.
• Collaborators included researchers from the University of Jena in Germany, highlighting international contributions to understanding AIs limitations in computational science.
• The findings raise concerns about deploying AI in research environments without oversight, as the lack of reasoning capabilities could lead to erroneous conclusions.