Anthropic are now actively using the approach to alignment often called “Alignment Pretraining” or “Safety Pretraining” — using Stochastic Gradient Descent on a large body of natural or synthetic documents showing the AI assistant doing the right thing. They tried this out, ound it works well, and are now using it.

I’m absolutely delighted. I’ve been advocating this approach on LessWrong and the Alignment Forum for several years:

• How to Control an LLM's Behavior (why my P(DOOM) went down)
• Motivating Alignment of LLM-Powered Agents: Easy for AGI, Hard for ASI?
• A "Bitter Lesson" Approach to Aligning AGI and ASI
• Why Aligning an LLM is Hard, and How to Make it Easier
• The Best Way to Align an LLM: Is Inner Alignment Now a Solved Problem?
• Motivating Alignment of LLM-Powered Agents: Easy for AGI, Hard for ASI?
• Pretraining on Aligned AI Data Dramatically Reduces Misalignment—Even After Post-Training

I’ve been very excited about this alignment technique for a couple of years, ever since I read the seminal paper demonstrating that it was extremely effective, Pretraining Language Models with Human Preferences (Korbak et al., ’23). This was later followed up by Safety Pretraining: Toward the Next Generation [...]

---

First published:

Source:

Linkpost URL:
https://www.anthropic.com/research/teaching-claude-why

---

Narrated by TYPE III AUDIO.