Christopher Berry
@cjpberry
Perpetually mystified | Views are mine.
Full house for a talk on AI Control. Roughly a third of Toronto’s AIS community is here tonight. Live stream soon: m.youtube.com/@Trajectory-...
Lewis Hammond is presenting some OC at Trajectory Labs tonight. Capabilities and Cooperation in Strategic Interactions. Alignment is insufficient!
Banger of a talk from Bruce Schneier tonight at Trajectory. www.schneier.com
We’re at Papers In The Park S2E11. A perfect day for it. Feels 21c with the gentlest wind and no mug. Thanks to @cohereforai.bsky.social for sponsoring, and Alvin for the walk through. Papers after the jump.
One of my favourite researchers, Rauno Arike from Aether is giving a talk. Gould, D., Ward, F. R., Woodruff, A. C., Arike, R., Hills, J., Serrano, A., ... & Stastny, J. (2026). Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models. arXiv preprint arXiv:2606.07157.
What a spectacular day for running on the Martin Goodman Trail! The geese feast under blue skies and better air.
I’m at Trajectory Labs tonight, where David is giving a talk on Recursive Self Improvement (RSI) tonight.
Papers In The Park S2E10. We’re covering the recent J-Space paper. Too long to print so there’s a summary. Thanks to @cohereforai.bsky.social and Alvin and Anthony for sponsoring!
Kristy, based at MATS, is giving a talk on US-China AI Relations at Trajectory Labs tonight!
It’s Papers In The Park! On a Sunday. (On account of the big game yesterday). Thanks to Alvin and Anthony for organizing, and to @cohereforai.bsky.social for sponsoring.
I’m enjoying this. Well written. Credible retelling of some Levin. Good stuff.
I’m at Papers In The Park! All the good people are here. We’re covering CALIBER: Calibrating Confidence Before and After Reasoning in Language Models. Thanks to Alvin and @cohereforai.bsky.social for sponsoring… and the paper! Shoutout to Finlay, Kurien, Dash, Fadaee, and Ermis!
It’s a beautiful day, 22c, sunny, for some papers in the park! Next-Latent Prediction Transformers Learn Compact World Models. Thanks to Alvin and @cohereforai.bsky.social for sponsoring!
Giles is asking if we can be mean to AI at tonight’s Trajectory Labs technical talk.
Ooooh, people in England and Croatia will have an opinion about this.
It’s S4E4 of Papers In The Park. Javad Rajabi is here to walk us through his paper on SEGA. arxiv.org/abs/2605.22668 Thanks to @cohereforai.bsky.social for sponsoring, to Anthony and Alvin for organizing, and Javad for walking us.
The SkyDome is open. Finally a beautiful day. I’m at the BDC this evening for the DevTools meetup.
Papers In The Park is at New Stadium today. Because of rain. We’re covering: arxiv.org/abs/2604.18519
“Put another way: Did maritime insurance in England cover the premeditated killing of Africans?”
“And once shared purpose disappears, it’s gone. A flaw lurked right at the core of the experiment, as flaws so often do in works of ambitious genius.”
“Against the Platonic Representation Hypothesis” from Robert Adragna. Thanks to Trajectory Labs for organizing! These are so good.
S2E2 of Papers In The Park! This week: DFlash: Block Diffusion for Flash Speculative Decoding arxiv.org/abs/2602.06036