AIs like Microsoft’s Copilot, Google’s Gemini and OpenAI’s ChatGPT still make serious errors rendering structural formulae
Two chemists are warning that using generative artificial intelligence (GAI) to produce chemistry-related images, such as molecular structures, is leading to serious errors and could end up damaging the education of the next generation of scientists.
Audrey Moores, a chemist specialising in nanoparticles, catalysis and green chemistry at McGill University, Canada, and Vânia Zuin Zeidler, a sustainability chemist at Leuphana University, Germany, joined forces to write a comment piece in Nature Reviews Chemistry in which they called on the chemistry community to ‘ban the use of GAI for any molecule or element representations’ after seeing numerous incorrect structures appearing in presentations and conference materials.