Intentional Control of Internal States in Gemma 3 27B

·LessWrong··

This research was done as my capstone project during ARBOx4.Epistemic Status: I'm relatively sure the results I obtained and my interpretations are correct. I'm unsure if the effect would replicate in a different setting and how much it differs between models.SummaryI replicated the Intentional Control of Internal States section of Anthropic's Emergent Introspective Awareness in Large Language Models (Lindsey, 2025) on Gemma 3 27B Instruct and found the same effect with smaller strength. When to...

Read full article →

Related Articles

Document-borne AI worms can self-propagate through Copilot for Word
Canopy9560 · Hacker News · 13h ago
AI's top startups are barely publishing their research
YeGoblynQueenne · Hacker News · 4h ago
Handbook.md shows that long policy documents do not reliably govern agents
spIrr · Hacker News · 12h ago
Keychron announces first open-source firmware for gaming mice
JLO64 · Hacker News · 8h ago
Turning a dumb AC unit smart (without losing my security deposit)
austinallegro · Hacker News · 7h ago