Loading activity…
@tallinzen
Professor @nyuling and @NYUDataScience, research scientist @GoogleAI, inventor of the word "bertology"
AI
Google DeepMind's expression-of-interest form describes a planned 24-week, in-person NYC project for PhD students starting in early 2027.
AI
A user says the figure comes from a talk they gave the previous day on studying introspection in LLMs.
AI
Arora shared defense slides titled “Causal Interpretability for Analysing and Improving Language Model Architectures.”
AI
One commentator says existential-risk framing can attract talent and coordinate research, even when a study’s connection to risk is indirect.
AI
A user mentions “harder things to RL on” and questions what would make Google “finally get serious.”
AI
A user pairs the joke about talk titles with a photo link.
AI
A user points to rising submissions in asking whether the conference should make such a substantial cut.
AI
A user says it seems good for more people to consider what they call a fundamental transformation in human history.
AI
A post describes “science again” as a common refrain at COLM, while questioning whether RL loss tweaks are better.
AI
A post argues advisors learn to track each student's project and next steps amid office visits every 30–60 minutes.
AI
A post describes Brown's remarks as “insightful.”
AI
A user suggests the voice for “maximum comedic effect,” sharing links to Scott Aaronson’s blog and a photo.
AI
A writer argues that agnosticism could reinforce existing prejudices and deepen epistemic closure amid uncertainty about AI consciousness.
AI
A contributor said Carter would present it at 4 p.m. on October 7 and later shared a paper link.
AI
Eric Zelikman argues regex-based and statistical signals can help, but should not become headline results.
AI
One commenter argues EAs are far more powerful, but far less coordinated, than people expect.
AI
A post says one of the paper’s collaborators will present it at 11 a.m.; the same user later links to an NYU CDS blog write-up.
AI
A commentator argues that assumed duties toward conscious AI make it harder to examine whether AI is conscious.
AI
A post invites questions at COLM during the week of October 5 about whether AI models can examine their own computations.
AI
Addressing artists at the Vatican, the pope called for a renewed alliance to protect human creativity.
