My friends all hate AI; I just joined an AI startup

The author faces pushback from friends and colleagues who are strongly anti-AI, making it awkward to reveal her return to AI in education. She acknowledges that much of the criticism is justified: the internet is oversaturated with AI-generated 'slop,' executives make overhyped claims verging on fraud, and people are outsourcing their thinking to AI. She warns that skills atrophy when unused, and that reading, writing, and understanding texts are core parts of being human.

Concrete problems abound: professor friends report students submitting AI-generated essays and reading scripts for presentations, and open-source maintainers are flooded with low-quality AI pull requests. Most AI education products are 'terrible,' chasing gameable metrics—one school uses AI-written passages and AI-generated quizzes instead of real novels, while another product scammed the Los Angeles Unified School District out of $3 million before the founder was charged with fraud and identity theft.

The author reflects on a decade of worry about AI. In 2016, she and Jeremy Howard co-founded fast.ai to counter the homogeneous elite controlling AI development, but today a handful of billionaires at top AI labs hold more power than ever. She gave 2018 talks critical of the labs' assumption of limitless computing power, and fast.ai won a 2018 Stanford competition against Google and Intel. She also founded the Center for Applied Data Ethics at the University of San Francisco to make ethics a core part of data science training.