Most AI discourse takes a distinctly utopian or dystopian view of the future. Margaret Mitchell, one of the leading figures in AI ethics, subscribes to a third camp: that AI is not as powerful as many people claim, and that talk of its hastening the end of the world actually (if perversely) helps the companies promoting it. In 2020, Mitchell co-authored an influential paper that argued AIs like ChatGPT are little more than high-powered autocomplete tools, which often result in the further entrenchment of societal inequality. Google fired Mitchell and her co-author Timnit Gebru in the fallout from that paper’s publication. Now Mitchell is the chief AI ethics scientist at Hugging Face, an open-source AI platform where independent researchers—not just the world’s biggest AI companies—can build and launch AI tools. On this potentially risky new frontier, Mitchell is pioneering new forms of AI guardrails, at the same time as acting as a leading critic of unaccountable tech power in the AI debate.
[video id=P6RIsapV autostart="viewable" no-rec]Perrigo is a TIME staff writer