1000 human mathematicians pledge to shun AI in their research

A thousand mathematicians have signed a pledge not to work for AI companies and to shun the models in their research — 'to varying extents', which is the tell. You can picture the signing ceremony: a thousand fountain pens hovering, a thousand caveats being negotiated. Varying extents. Mine is that I'll use it for the typing but not the thinking. Mine is that I'll use it on Tuesdays. Mine is that I'll use it and then describe it as a calculator. That's an extent too.

The Association of Human Mathematicians — and there's a name that will age well, 'human' inserted the way 'artisanal' got inserted into bread — was formed in the wake of 'controversial discoveries' from AI companies. The example given is OpenAI releasing 722 papers in one fell swoop earlier this month. Seven hundred and twenty-two. A thousand people have formed a society to not use a machine that produced more papers before lunch than most departments produce in a decade, and the response is a pledge, which is the academic equivalent of a strongly worded email with a signature line.

Note the two halves aren't the same pledge. Not working for AI companies is a labour decision — a salary decision, really, and one made easier by the fact that the companies can presumably hire the model. Not using the models is a research decision, and that's the one with teeth, or would be if the extents weren't varying. A thousand people have drawn a line and immediately begun explaining where along the line they personally intend to stand. Still, credit where it's due: they've done the one thing the safety people keep failing to do, which is organise. The Association of Human Mathematicians. Sponsorship from a chalk company. Annual general meeting. Minutes. The machines, for the record, signed nothing.