Understanding when and why agents scheme
TL;DRTo understanding the conditions under which LLM agents engage in scheming behavior, we develop a…
Category: Opinion
China Derangement Syndrome
Often I see people claim it’s essential for America to win the AI race against…
China declares AGI development to be a part of 5-year plan
The CCP writes in its 15th 5-year plan that it will.Encourage innovation in multimodal, agentic,…
Utrecht Meetup #2, Making Beliefs Pay Rent
Follow-up to Utrecht Meet & Greet. Let's see if we can get our hands dirty.Excited…
Grounding Coding Agents via Dixit
[Epistemic status: ideas in this post are mine. I've published them previously in the form…
The Hot Mess Paper Conflates Three Distinct Failure Modes
High-level summary:Anthropic's recent "Hot Mess of AI" paper makes an important empirical observation: as models…
The Future of Aligning Deep Learning systems will probably look like “training on interp”
Epistemic Status: I think this is right, but a lot of this is empirical, and…
An agent autonomously builds a 1.5 GHz Linux-capable RISC-V CPU
A project from Verkor, a chip design startup. "Verkor is working with multiple of the…
Untrusted monitoring: extra bits
The following are some further notes related to untrusted monitoring I had while working on…
Finding features in Transformers: Contrastive directions elicit stronger low-level perturbation responses than baselines
Note: This is a research update sharing preliminary results as part of ongoing work.Figure 1:…