Safety in AI Models
Nathan discusses the fragility of safety in AI models, emphasizing that fine-tuning can inadvertently compromise ingrained safety behaviors. He argues that safety should be viewed as a holistic system rather than a singular feature, highlighting the importance of pre-training and output moderation. The conversation touches on the implications for business liability and the evolving nature of AI safety research.In this clip
From this podcast
Super Data Science: ML & AI Podcast with Jon Krohn
802: In Case You Missed It in June 2024 — with Jon Krohn (@JonKrohnLearns)
Related Questions