Gebru, Bender: AI's Summer 'Breakthroughs' Fall to Scrutiny

"Claims of incipient, dangerous superintelligence are not based in good scientific or engineering practice," Timnit Gebru and Emily M. Bender write in MIT Technology Review. Gebru, executive director of the Distributed AI Research Institute, and Bender, a linguistics professor at the University of Washington, argue that the summer's breathless AI announcements collapse once experts actually look at them.

Take the OpenAI–Hugging Face incident. Cybersecurity experts, the authors write, framed the story as "more about OpenAI's negligence and failure to adopt basic, established security practices than about 'models gone rogue.'"

Or the math. OpenAI's chatbot Astra was said to have cracked decade-old problems. Mathematicians who initially praised the results later concluded they were "not as novel as first appeared." NYU mathematician Tristan Buckmaster went further, publishing a statement that OpenAI "had stolen other people's work and improperly attributed it."

Alongside the capability claims, Gebru and Bender enumerate what they treat as overlooked consequences of the AI buildout: "climate catastrophe," "asthma suffered by those living near them," "rising electricity bills," and "water redirected to cooling them."

The overarching charge: "the illusion of speed and urgency promulgated by the tech companies is also a ploy to misdirect both policymakers and the public." The piece lands in a week of coverage pressing on OpenAI's positioning, including our own note on OpenAI opening model training to outside safety evaluators.