Working on math AI at acornprover.org and telescope software at deepsynoptic.org. Formerly: Parse cofounder, Facebook, Google
Joined March 2008
- Tweets 18,399
- Following 1,458
- Followers 7,149
- Likes 22,790
704 Photos and videos
Pinned Tweet
28 Jan 2025
I'm happy to announce the launch of Acorn, a new theorem prover that includes an integrated AI.
Theorem provers let you write mathematical proofs that are rigorously verified. But they are notoriously difficult to use. Acorn makes it easier, by using AI to fill in the details.
9
14
104
16,866
Jun 12
wife: So did she accept his proposal?
me: Ah, it didn't come up. We just talked about the Knicks
1
275
Jun 11
Wrote a post about philosophy vs engineering, whether AI can really "do math" or "make art". open.substack.com/pub/lacker…
1
7
613
Jun 10
I feel like over time the Claudes speak more and more in their own funny, nonlinear dialect, full of metaphor and abbreviation.
“If they all land, you have rung ii locked (the conditional fully written) and rung i at the mercy of nothing but the gluing.”
4
1
11
2,037
Jun 10
Reminds me of when Zuck started publicly advocating for a more pro-immigration policy, back in 2013. I remember thinking, he has some good points, I wish politicians would wake up and pay attention to the immigration rules. Then the monkey's paw curled... darioamodei.com/post/policy-…
5
235
Jun 10
The great thing about the AI boom is that there are no monopolies, no gatekeepers. No Wintel, no App Store. No lockin, not really, not yet anyway. I hope Anthropic and OpenAI both stay at the frontier, keeping each other honest, for a long time to come. Rooting for Google too.
14
7
119
26,686
Jun 10
Fairly weird behavior from Fable. It’s almost begging me to download a bunch of files it created, and re-upload them, in order to change its own context. It claims it’ll be much more efficient at problem-solving if I do!
2
5
4,491
Jun 9
fable experience so far: it thinks hard, takes up 10% of my quota, and then falls back to opus, claiming it isn't safe to let fable solve my math problem
1
3
289
Jun 8
I think I will try to write a Substack. I promise nothing. There's a decent change I will fail to write anything. But if you like my tweets, why not subscribe? lacker.substack.com/p/hello-…
1
5
382
Jun 7
At a bookstore, looking through the science fiction section. Despite the “futurism” vibe, not a single book grapples with modern AI. It’s like the whole genre exists in an alternative history, where AI never happens.
33
71
5,618
Jun 3
Entertained by professor Garcia here.
1. Claims he is a "strong, strong opponent" of curving a class so that only a limited number of students can get A's
2. Fails 35% of the students in CS 10, saying they all deserve to fail
dailycal.org/news/campus/aca…
1
4
836
Jun 3
I liked this piece... but I felt like it dodged the most interesting questions, by focusing on "is AI conscious".
What would it be like if AI became better at writing than humans?
I'll admit, it's quite hard for me to imagine that world. But I think Ted Chiang could imagine it.
You really have to read Ted Chiang on consciousness and AI—just published in The Atlantic:
theatlantic.com/philosophy/2…
6
5
571
Jun 2
A really interesting document. My takeaways:
1. the UC student body is about:
20% Hispanic
50% Asian
25% white
5% other
2. Despite all the arguments, laws, and rule changes, that only changed a few % over the past 10 years
Jun 2
One motivation to ban SAT/ACT in UC admissions was to eliminate perceived bias against black and latino students.
But the biggest shift by banning the SAT was a transfer of admit seats from Asian students to white: docs.google.com/spreadsheets…
E->F, rows 12-19.
2
8
3,483
Jun 1
I wonder if there will be fun video games that require an RTX Spark to play.
SITUATION DETECTED: Nvidia has announced RTX Spark, its first consumer chip for Windows laptops and desktops.
The ARM-based superchip combines a 20-core Grace CPU, Blackwell GPU, and 128GB unified memory in a single package. Coming Fall 2026.
2
298
May 29
thoughts on modern software.
the cost of debugging "normal algorithms" is now incredibly low. but the cost of debugging "weird algorithms" is still fairly high.
perhaps you should try to boil the ocean with "normal features" but restrain yourself on the "weird features".
341
May 27
Soon the math journals are going to get overwhelmed by AI submissions, like many open source GitHub repos have.
May 27
1/ Since February, 8 papers across algebraic geometry, representation theory, number theory, combinatorics have been quietly appearing on arXiv.
Proofs by AxiomProver.
5 papers are now accepted at solid peer-reviewed math journals. To our knowledge, a first for the literature.
4
2
36
7,598
Kevin Lacker retweeted
May 22
Prominent mathematicians as recently as the 17th century were proving amazing things, yet were still uncomfortable with negative numbers.
Cardano called them “fictitious.”
Pascal said that subtracting 4 from 0 is “utter nonsense.”
Descartes had mixed feelings but called negative roots “false.”
Leibniz said they are just a tools to guide one’s thinking, or an indication that the question with poorly formed.
May 12
What historical fact sounds fake but is true?
32
32
493
53,778
May 22
The bottleneck to software engineering is now reading long explanations that are 80% correct, 20% confused, from the AIs
1
230