steerable and explainable AI
San Francisco
Joined June 2024
- Tweets33
- Following34
- Followers296
- Likes182
We're sponsoring and attending this :) see you there!
Metagame is BACK!
Three days of games, puzzles, workshops, talks, competitions and things that don't quite fit into any of those categories.
November 6-8 at Lighthaven in Berkeley.
Early Bird pricing here September 30, so get your tickets now!
metagame.games
d_model retweeted
dmodel is hosting a tech week party! if you're interested in interpretability or worried about alignment and looking to discuss solutions with other researchers in the field please join us next thursday in soma!
limited space, more info here:
luma.com/vrq4xx25
We're hosting an after-party at Spark Social SF for @ycombinator Startup School 2026 builders! Come grab some drinks + snacks and meet the d_model team.
📍 Spark Social SF
🗓️ Saturday, July 25 | 6–8pm PT
🎟️ Register → luma.com/jz0h1t7q
Registration is on a first come, first serve basis. Everyone who registers is entered into our raffle with the chance to win gift cards and d_model swag. We'll draw winners at the event, so show up for your chance to win!
d_model retweeted
We're working directly with frontier-lab researchers to do this work with bleeding-edge models.
We expect to produce many more blog posts like this as our research program grows symbiotically with our RL-env work.
And we're hiring :)
dmodel.ai
Discovering Concept-Editing Algorithms With LLM Agents dmodel.ai/concept-erasure/
d_model retweeted
I've shifted my research to focus on automated alignment research. We will have automated AI research very soon and it's important that alignment can keep up during the intelligence explosion.
d_model retweeted
I'm doing research @d_model_ai now! Very excited about the team and mission. We're teaching computers to teach computers to teach computers to be good people.
d_model retweeted
the true value in AI coding nowadays isn’t in writing code, it’s in reading code: outlining large code bases, navigating, retrieving suggestions & ideas, and breaking down and explaining tricky concepts
i’ve never had a more complete understanding of codebases before
I am learning that I am an unusual CEO in many ways. Some know that I still code on nights and weekends (github.com/lattner) despite a busy “day job”.
As part of that, I care deeply about sw development as a profession, and what “AI anxiety” is doing to many early and experienced developers.
IMO, AI is here to stay, super useful today for things, and will continue to improve… but is far from SWE job displacement. It is sad for me to see many talented people giving up hope and sabotaging their own career development because of a future that may not actually arrive!
Please watch the video for a much more nuanced discussion about the issues involved, particularly if you are an early career engineer who wants to make a big impact on things!!
d_model retweeted
📢 Can LLMs really reason outside the box in math? Or are they just remixing familiar strategies?
Remember DeepSeek R1, o1 have impressed us on Olympiad-level math but also they were failing at simple arithmetic 😬
We built a benchmark to find out → OMEGA Ω 📐
💥 We found that although very powerful, RL struggles to compose skills and to innovate new strategies that were not seen during training. 👇
work w. @UCBerkeley @allen_ai
A thread on what we learned 🧵
