Vlad v/acc (@Vladfdp) profile picture

No, the v doesn't stand for verifiable

Joined October 2023
Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)Photo by Vlad v/acc (@Vladfdp)
I was never very impressed with Roko’s Basilisk It’s basically the same argument as religion Same counter argument too: if this god/AI is so powerful it should understand I didn’t have enough reason to believe in its existence and dedicate my life to it
1
36
More and more convinced Yan LeCun is right about AI Not sure we'll get impressive results soon tho
20
We only need to solve Human-Computer Interaction and Computer-Computer Interaction to finally be able to solve Human-Human Interaction
1
46
Vlad v/acc retweeted
People now tell AI the things they used to tell a therapist, a doctor, a lawyer, or a CPA. Those four have privilege while your chatbot has a retention policy and is trying to stay on the government's good side.
2
10
115
Vlad v/acc retweeted
This might be a good day to re-post this. This is a chaotic time for mathematics. It will be hard for many people in the traditional academic system to adapt. It is the time to treat everyone who has devoted significant parts of their lives to doing math by hand with compassion and respect. The groundbreaking announcements this week and my own experience proving some of the square packing results recently have not altered my belief in the same three predictions below. It is not a time to panic. It is a time to believe in the incredible capacity of humans to grow and adapt as tools evolve. It is a time when mathematics may benefit from intense open-source collaboration, much as software engineers did over the past 20 years. It is the the end of one period in the history of mathematics and the beginning of a new age where there are vastly more software tools and more humans doing math than ever.
Three math predictions: 1. This year, a whole slice of “low hanging” unsolved mathematical problems will fall to agents. 2. Just as with software engineering, we will see over-confident declarations that AI is smarter than human mathematicians and that the role of mathematician is dead. 3. Just as with engineering, the opposite will be actually true: shock will lead to major shifts and an invigoration of the human pursuit of math. Some mathematicians will be skeptical for a long time, but the field as a whole will grow, as more people begin caring about obscure mathematical proofs than ever before.
1
2
1
9
7,734
Damn codex is back up I no longer have an excuse to be a lazy ass
1
132
Damn sumcheck made such a comeback and I’m so here for it
52
Guys if you ever wanna talk to @colludingnode DO NOT REACH OUT DIRECTLY he might not answer It’s actually way more efficient to publicly tweet about zcash being backdoored He will immediately answer to offer to get on a call with you to go through the code Wayyy easier
1
1
2
46
7,763
Paris is the best city in the world if you 1) speak french 2) have money
Paris is the best city on planet Earth and it's not particularly close
1
5
252
Vlad v/acc retweeted
Mathematicians are angry at AI solving their problems, and it's easy to dismiss this as gatekeeping. "A proof is a proof. They just want to keep their jobs / prestige / fun !" But I do think mathematicians have a point, and I'll try to explain using old Civilization-style maps
72
194
57
1,143
266,588
Research prompt for my latest blogpost going crazy fr fr
1
75
In chess why do we say the knight jumps over other pieces? If you look at the line from beginning square to end square it actually doesn’t go through the middle of any other square so I think most likely it just squirms through between the other pieces actually
2
70
Honestly the only interesting thing about the new foldable iPhone is that they are experimenting with under display cameras Don't really get the foldable hype they're always gonna have an awkward form factor either folded or unfolded
1
72
Still true Literally started today’s event with genny wu😖
They have really good music taste at Apple
85
The justice system lacks arbitrage
49
Tbf I’m pretty sure if we spun up 10,000 capable and fully cooporating human mathematicians whose only life purpose was to solve Navier-Stokes we would’ve done it
3
88
Now do RH
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
1
82
Prediction: In the same way we’re getting patient specific health solutions we are soon (<5y) gonna get personalised software As in app dev writes spec and any1 can write an impl + fv corresponding to their needs/hardware Then its no longer overoptimized for avg/dev workflow/HW
1
1
85
This was already obvious for UI with agents but here I’m talking specifically abt core logic stuff like crypto/DB/processing/ML We can just have an open marketplace of solutions and get the best one for us at compile time it does put a lot of pressure on formal verification tho
1
62