The LLM warnings Google fired Timnit Gebru over have all come true

Timnit Gebru was fired from Google in December 2020 for refusing to retract a research paper warning about dangers of large language models, and all four major predictions in that paper have since come true: hallucination problems, bias amplification in deployed systems, massive environmental costs, and documentation gaps that enabled training on harmful data. The paper predicted that scaling up language models would produce systems appearing intelligent but lacking comprehension, would amplify biases in hiring and healthcare decisions, would create massive carbon emissions, and would make datasets impossible to audit—outcomes now extensively documented across the industry. Google publicly claimed she resigned while Gebru and over 2,600 colleagues confirmed she was fired by email while on vacation for refusing to compromise on the research.