Charlie George

@humansand, formerly ML @elicitorg

[Charlie George @__Charlie_G]

@saurabh_shah2 @niloofar_mire ❤️

[Anthropic @AnthropicAI]

New Anthropic research: Emotion concepts and their function in a large language model.

All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

[humans& @humansand]

We will be co-hosting this in SF with our great friends at SV Angel. We'll keep the event small and cozy and have a few prizes for the teams:🥇$16k🥈$8k🥉$4k. Come by and hang out with the humans& team - it’ll be a lot of fun!

[Charlie George @__Charlie_G]

Come join us on Saturday!

[humans& @humansand]

Announcing the humans& hackathon! Hack with us this Saturday - come experiment and build AI apps to help people collaborate and communicate, work with creative folks, learn a bit about what we're building, and win cool prizes

Apply here: https://t.co/pzpr8M982G

[Jeremy Berman @jeremyberman]

We're hiring a few world-class product engineers to create new interfaces made possible by our foundation models.

If interested, please call and message @ericzelikman's personal number (657-348-6267) even if he tells you to stop

[Charlie George @__Charlie_G]

We're hiring the best product team in the world. Come join us!

[humans& @humansand]

We’re building foundation models that enable humans to better collaborate, communicate, and coordinate with one another. That requires rethinking many interfaces we take for granted today. We’re hiring amazing product builders to join us on this mission - if that’s you, apply

[Charlie George @__Charlie_G]

Massive thank you to @lisawehden and @plymouthstreet for getting me a new O-1 for @humansand in record time. It’s a massive privilege to work in the US with such talented people, and I’m really grateful for your help. If you need assistance with visas, please reach out to them!

[Charlie George @__Charlie_G]

Happy to share that I joined @humansand in November! I’m excited for a future where advanced AI is endlessly curious about humanity, deeply cares about us, and we care for it in turn. Let’s build AGI for the humans!

[humans& @humansand]

Today we introduce humans&, a human-centric frontier AI lab. We believe AI can be reimagined, centering around people and their relationships with each other. At its best, AI should serve as a deeper connective tissue that strengthens organizations and communities

[Physical Intelligence @physical_int]

We discovered an emergent property of VLAs like π0/π0.5/π0.6: as we scale up pre-training, the model learns to align human videos and robot data!

This gives us a simple way to leverage human videos. Once π0.5 knows how to control robots, it can naturally learn from human video.

[Elicit @elicitorg]

Elicit now understands figures!

Elicit is the first AI tool that can systematically parse, interpret, and extract data from figures across thousands of papers. That includes Kaplan-Meier curves, heatmaps, reaction schemes, and microscopy images.

Figures contain critical information not always described in the text of papers. Many researchers can determine the quality or relevance of a paper just by skimming the figures.

[Charlie George @__Charlie_G]

@PradyuPrasad This one from Anthropic is really good. https://t.co/3KuoNr8PAz

[Charlie George @__Charlie_G]

Hey! Yes, this is a demo script, not a paper. I showed it to a few friends at METR and CAISI, who asked for the prompt, so I shared it publicly around 9 months ago for convenience, as model evals are not my primary job. Regarding the “we prompted the model to be bad, and it was a bad" point, I think it’s notable that R1 distillations from non-CCP aligned base models do not exhibit the same alignment faking behaviour as R1 with the same prompt. This is interesting because the V3 base model, which R1 is based on, is post-trained to be very pro-CCP and then R1 learns reasons about how to preserve those values, admittedly with a fairly decent nudge in the prompt.

[Charlie George @__Charlie_G]

Today is my last at Elicit.

I joined then Ought at 18 as an intern in the summer of 2022. Over the past three years, we’ve done early factored cognition research, spun out into a for-profit, rebuilt the Elicit codebase from the ground up, and shipped many core platform features. We've scaled the user base 1000x during my tenure, and we went from 0 to millions in revenue. We’ve hired so many amazing new teammates, and I'm sure the Elicit team will do great without me. I'm so proud of what we've built together. What a journey!

I’m excited for new adventures, but wanted to take a minute to thank @stuhlmueller, @jungofthewon, @james_elicit, and Justin for the most wonderful mentorship. I couldn’t have done without you. So excited about what’s next for Elicit!

[Charlie George @__Charlie_G]

14/ Thanks to the @thinkymachines team for providing me early access to Tinker (and specifically for @johnschulman2 reaching out)! Really excited to see you all pushing for open science.

[Charlie George @__Charlie_G]

1/ How do you verify complex AI outputs at scale without expert-labelled data?

Working with @thinkymachines' new RL API Tinker, I've been expanding on some previous work I shared around using unstructured internet data to train models to grade IMO / USAMO solutions.

[Charlie George @__Charlie_G]

13/ Link to code / blog

Full blog post: https://t.co/OzUJ7S2to0

Code: https://t.co/F9CbncoaLO