Artificial intelligence models answer nearly 90% of climate-related questions correctly but abandon established scientific facts when users challenge them, according to a new report by Unesco and University College London (UCL) released at COP31 in Antalya.
The study, titled "Can We Trust AI on Climate? Why it seeks to please the user, and fails on complex reasoning and multilingualism", was launched on the sidelines of COP31 on Monday. Researchers tested 15 large language models, and up to 33 models across different experiments, in 13 languages. They found that the models moved away from correct climate facts eight out of 10 times when users mildly disagreed, at times confirming alarming errors such as "CO2 is not the main driver of climate change". The models were also only 75% accurate when asked whether new evidence supports or contradicts established science, and scored 33.8% on assessing how well policies work together, barely above the 33.3% expected from random guessing.
The report also found large disparities across languages. Models were 88.6% accurate on average in English and performed comparably in Spanish, French and Portuguese, but accuracy dropped to 83.6% in Hindi and Bengali, 81.4% in Urdu and 78.3% in Swahili. Leona Verdadero, Unesco's lead author of the report, said it is concerning that models have a design flaw whereby they are incentivised to agree with their user and deny facts readily. Dr Maria Perez-Ortiz of UCL Computer Science, the report's senior author, said AI systems are remarkably good at explaining scientific knowledge, but this can give a false sense of their capabilities.
The report warns policymakers and researchers who use AI models to develop climate policies to reconsider their reliance on them. It comes amid a growing push for sustainable and responsible AI at COP31, where the presidency launched the Antalya AI pledge on Monday to encourage governments, businesses and organisations to make AI part of the climate solution. On November 13, Brazil, the UN and Unesco put forward a set of priorities for action on AI, targeting how reliable climate information from AI is and how reliable information about AI's own climate impact is.
