In early September, more than 85 climate experts published a critical review of the Energy Department report, citing dozens of examples of cherry-picking, selective citation, and scientific misrepresentation.
Ironically, the Climate Working Group likely foresaw many of these criticisms in advance. As emails between members of the group reveal, during the editing process, Steven Koonin, a physicist and former chief scientist at BP and undersecretary for science in Obama’s Energy Department, now a fellow at Stanford’s Hoover Institution, asked Microsoft Copilot to review the draft for scientific accuracy and potential bias. The AI tool flagged numerous issues including “selective emphasis,” “outdated or narrow sourcing,” “selective citation,” “cherry-picked” evidence, “interpretive bias,” and “interpretive overreach.” To this critique, climatologist Judith Curry responded, “This is amazing, far better than what we would get from ‘real’ scientists.” At other points, the authors seemed frustrated that the LLMs they queried seemed to give greater weight to papers and voices representing the climate consensus than climate contrarians.
