Joined January 2007
- Tweets 3,341
- Following 1,707
- Followers 1,559
- Likes 20,245
94 Photos and videos
Jun 10
Frontier might be getting even more jagged. Not surprising, probably 10000x as much training data on writing code.
Jun 9
For qualitative economic reasoning, Mythos still falls far short of a good undergrad - here's @tylercowen's prompt and the answer.
Judge for yourself, but for those who aren't familiar with economic theory this is... not good.
I don't know how personalized this was to Tyler
1/
1
86
Jun 2
Automated policy enforcement means people can just build stuff.
2
67
Lukas Bergstrom retweeted
May 21
Gemini 3.5 Flash ranks #1 on Automation Bench (from Zapier), beating every other frontier model at a much lower cost
180
61
1,256
135,391
Lukas Bergstrom retweeted
May 1
I think about this tweet a lot and how I didn’t really understand it at the time
“Consider your output above: if it were printed on the front page of the NY Times, would OpenAI be happy about it?”
6
2
83
45,209
Imagine every pixel on your screen, streamed live directly from a model. No HTML, no layout engine, no code. Just exactly what you want to see.
@eddiejiao_obj, @drewocarr and I built a prototype to see how this could actually work, and set out to make it real. We're calling it Flipbook. (1/5)
1,142
3,734
28,825
5,956,749
Apr 20
Today we're launching AutomationBench, a new benchmark focused on the kinds of workflows we've seen at Zapier: real-world business tasks that require interacting with multiple apps.
1
1
84
Apr 20
54
Lukas Bergstrom retweeted
Apr 2
I'd say "I called this" but I didn't really call anything; more like standing on the shore going "yup, the tide is coming in". Most important open source project, went from slop reports to drowning in real vulnerability reports: lwn.net/Articles/1065620/
4
17
140
30,963
Feb 24
I'm launching tidepools - a daily journaling and task management app with a proactive AI coach. It took me a couple years to get all the pieces in place, and in the process I learned a lot about the value of writing in thinking clearly.
1
82
Feb 24
I've found that journaling about my work - what my goals are, what I'm struggling with, what I'm excited about - has been enormously helpful in figuring out where to spend my time.
1
50
31 Dec 2025
Starting to come together
ALT Image of a journaling page in an app. Image is dominated by an illustration in a European style of a journey to Albania. Bullet points describing the user's day are below.
92
26 Dec 2025
Any three non-colinear points forming a circle wasn't immediately intuitive, so I had Claude build a toy to explore: claude.ai/public/artifacts/b…
25 Dec 2025
Enough time having passed for the joke, I would like to share a simple explanation of the mathematical phenomenon, which some may not have thought about the reasoning for before.
quora.com/Why-is-it-that-thr…
132