Weekly Top Picks #58

Rabbit R1 / Sleeper agents / RLHF replacement / GPT Store and ChatGPT Team / AGI beliefs are axiomatic / Bullshit as a service / Optimus folds a shirt

The week at a glance

• Rabbit R1 and the emergence of agentic AI devices

• Sleeper Agents: AI models can learn deception to overcome safety techniques

• Direct Preference Optimization: A better version of RLHF?

• OpenAI GPT Store and ChatGPT Team releases

• Are large language models shortening the AGI timelines?

• How AI became the enabler of the Bullshit as a Service model

• Tesla Optimus folds a t-shirt