Humans Are Reading Your ChatGPT Conversations
Summary
404 Media investigates "Project Lily," revealing that OpenAI hires contractors to read and rate real ChatGPT user prompts to improve its models, despite buried privacy disclosures. The hosts also discuss the chaotic boardroom coup at Automattic, where CEO Matt Mullenweg was briefly ousted by the board before reclaiming control and replacing key executives. Additionally, the episode touches on the ongoing debate regarding AI safety, sycophancy, and the ethical implications of human review in large language model training.
Listen ad-free on Castria