Reactions: AI, Chemical Safety Board, and more

Regulate AI assistance in ACS PublicationsSince the somewhat sudden and progressive incorporation of artificial intelligence tools in scientific communication and education at all levels, both pros and cons have been repeatedly echoed. It is clearly debatable whether AI assistance in writing and reviewing, for instance, may or may not be regarded as fraudulent. However, to many of us who have transitioned from analog to digital ages, it would certainly be a pity to see the art of scientific writing end up in a gloomy, low-quality, and largely automatic activity. When recently I reviewed a long article for a well-reputed specialized journal, I began to suspect AI tools were behind it (there were disconnected paragraphs, some lacking sense and context, and numerous citations in text did not match those of the reference section). Beyond rhetorical speeches and lament, a few actions could be taken to avoid further threat to established standards. American Chemical Society journals, which are positioned among the most valuable and reputable publications, could introduce some preliminary initiatives. Consider, for instance, the inclusion of responsible declarations, explicitly identifying which parts of a manuscript, and to what extent, were elaborated, assisted, or assessed by language models. Such a statement adds to others, already regulated, like authors’ contributions, data availability, funding, or safety. A signed consent by all coauthors adhering to strict rules of use should also be compulsory before publication and will contribute to mitigating unethical practices. Proactive measures not only by publishers but also and above all by governments