Thanks, David. I like your attempt at fixing it. It reminded me of something @serkancabi and I tried a while back inspired by the awareness work of Michael Graziano, In addition to Bot and User, we added another party in the discussion: the Narrator. The Narrator generates a
RESEARCH
-
MedicBrain: AI Project Sharing Medical Knowledge for Cancer Care
By
–
Pour ça que je fais mon projet MedicBrain parcque j'ai fait le même constat avec mon enfant qui a eu le cancer, mais nous ils s'occupent très bien et leurs connaissances doivent être partagées.
-

NeurIPS Happy Hour: AI+Science and AI+Math Discussion
By
–
Join our second happy hour meetup today (Saturday) at @NeurIPSConf to chat about AI+Science and AI+Math with me and my team from @Caltech at: Achilles Coffee Roasters Gaslamp, San Diego. 5:15pm – 7:15pm We have a number of papers at NeurIPS on Neural Operators, Inverse
-
Why LLMs Require New Chats for Optimal Performance
By
–
Why is it that with ChatGPT, Gemini, Claude, Copilot and other LLMs we have to always start new chats for them to work well? What is the scientific explanation? What are the hypotheses? What is the evidence for each?
-
DeepSeek Model Distillation Continues Strong Momentum
By
–
Looks like the DeepSeek distillation is still going strong
-
Claude’s Post-Training Excellence and Alignment in AI Models
By
–
Every time you are chatting with claude(Opus 4.5 esp. but seems to apply to all models), you quickly realize how clever claude post-training is. The attention to details, data quality. It's rare that claude will break in the middle of conversations. I think this is beyond system prompt and training dynamics/physics, and more about aura-like things: behaviors, personalities, when to push back/when to defer, and countless subtle edge cases. It's no surprise alignment people there follow closely the training. There was an invited talk at CMU this past spring on alignment and I asked why claude vibes and post-training feel different, the answer was high-level as you'd expect, but same: training, post-training, alignment, and evals teams are closed-loop. Sam Bowman (@sleepinyourhat) From everything we know so far, Opus 4.5 seems to be the best-aligned model out there in a bunch of ways. I follow the training process closely as part of my work on alignment evaluations. Here's my guess about the two things that are most responsible for making 4.5 special. 🧵 — https://nitter.net/sleepinyourhat/status/1997006353647522098#m
-

Laude Institute Launches Open Frontier Conference for Open Research
By
–
Excited to see @LaudeInstitute announce Open Frontier — a new conference and collaboration for open research + open-source AI at the frontier. Proud to be a part this effort and excited to see all the collaborations that come out of it. https://
openfrontier.ai -

DeepSeek-V3.2: Open-Source Reasoning Model Rivals GPT-5
By
–
DeepSeek recently dropped a new open-source reasoning champion to rival GPT-5! Introducing DeepSeek-V3.2: 671B param Mixture-of-Experts (37B active) Gold-medal IMO & IOI performance Sparse attention for long contexts Balanced daily driver + tool-use Fully
-

NeurIPS Mexico City: Intimate AI Research Community Experience
By
–
Really enjoying the smaller scale and more intimate experience of NeurIPS in Mexico City, along with my @Mila_Quebec crew! Including dinners that feature guacamole with crispy larva 🙂

