Climate skeptics used AI to review their work. It told them they were biased (thebulletin.org)

A recent development reveals that climate skeptics selected by the Trump administration to review climate science findings employed AI large language models (LLMs) for their work, only to be confronted with evidence of bias in their arguments. Initially aimed at critiquing climate science, these experts utilized AI tools to assess their writing and acknowledged that their findings were often cherry-picked from narrow sources. A judge had previously ruled that the advisory group's formation violated the Federal Advisory Committee Act by lacking transparency and balance. Following this ruling, related documents were released, shedding light on how AI was woven into their workflow, serving various roles such as editor and pseudo-peer reviewer. The implications of this revelation are significant for the AI/ML community, highlighting the evolving intersection of AI tools and scientific research. While the group claimed not to rely heavily on AI for content creation, they consistently sought its input to enhance clarity and neutrality in their language. This case also underscores a critical tension within AI use; LLMs showed a preference for consensus viewpoints over contrarian positions, prompting frustrations from the climate skeptics who used them. Overall, this situation illustrates the nuanced role AI is beginning to play in academic research, emphasizing the need for careful integration of AI feedback to avoid bias while conducting scientific inquiry.