Professor @feup_porto and researcher @inesctec. Distributed Systems and Data. Co-creator of CRDTs. Still searching for unknown unknowns

Oporto, Portugal
Carlos Baquero retweeted
if you want to understand how AI will progress, study how bots changed chess and poker. Compare and contrast. the best humans can’t beat top bots at chess. Never. Poker is different. Bots beat pros over large samples, but a human can still sit down and win a session. That's because poker is a game of incomplete information and chess isn't. a bot can play close to game theory optimal poker, but the hidden cards and variance give humans a fighting chance, especially in the short run. Life is poker, not chess. It’s a game of incomplete information and luck. No amount of computing power eliminates that reality. This means the best bot won't be the one that wins every hand. It'll be the one with an edge that compounds over thousands of decisions. In the short run, anyone can win. In the long run, the edge gets paid. In chess, ‘human-plus-bot’ teams once beat bots alone, until the engines got too strong and the humans just got in the way. In a world full of hidden information, that partnership lasts longer. SO, in the future, the winner won't be human or bot. It'll be a human who knows what the bot can't see, when to use judgment and intervene, and when to shut up and let it play.
17
10
138
5,819
Carlos Baquero retweeted
AI Glasses make someone into a “Chess Grandmaster”. Te reality is nothing will go untouched by AI, for better or worse. Awareness is your friend.
Dealer Brandon
47
62
587
35,214
Preparing for an Informatics Europe session where I plan to address both the past publishing patterns and the AI transformation of research. Not an easy topic, since we are all figuring out how to cope with the exponential. Will probably frame it as questions for audience.
1
3
188
Carlos Baquero retweeted
Bitcoin's supply used to grow 50%+ a year. Today it's 0.82% 📉 Gold's: ~1.5-2% Bitcoin crossed BELOW gold in May 2024, and it's never coming back. - - - The schedule is already written: ~0.4% after 2028, ~0.2% after 2032, ~0.1% after 2036. When gold's price rises, miners dig more. When Bitcoin's price rises, supply doesn't budge. Hardest money ever made, people 🧱
9
22
230
7,652
Carlos Baquero retweeted
WHAT ??
958
13,510
266,162
8,827,396
Carlos Baquero retweeted
Have to give the EU regulators credit where it’s due for making all Apple products use USB-C
3,538
5,760
163,619
7,753,324
Carlos Baquero retweeted
You can be globally compassionate as a person and still feel that your society should have careful borders and that local cultures matter, and should be preserved if they're clearly doing well. These actually aren't in contradiction.
258
610
7,589
210,859
Carlos Baquero retweeted
"AI and Communications of the ACM," by Editor-in-Chief @JamesLarus, offers some promises and words of advice to authors who submit articles written with AI. bit.ly/4k0VxN1
2
6
525
Carlos Baquero retweeted
The tragedy of it all is that rioting against inflation and cost of living by destroying stores just increases inflation and cost of living. Fiscal issues lead to social issues, which lead to more fiscal issues and more social issues.
“We become what we behold.” -William Blake If you contemplate an archetype long enough, when the moment arises, you will unconsciously embody it.
263
302
3,417
182,432
I think model capabilities shouldn’t be judged on how hard maths look to a human regardless of how much an expert that person is. I think they should be judged in how “automatable” and brute-forceable said math problems are. A human taking 10 years to do something means nothing. A human may need ten years or more to run the US census while an IBM machine with perforated cards takes minutes.
In January, Jonathan Gorard (@getjonwithit) was confident AI couldn’t produce a machine-checked Lean proof of Maryna Viazovska’s Fields Medal sphere-packing result before late 2027. By March, one existed. He says the real story turned out to be complicated. Full episode: piped.video/oQypVVv1u1o
3
1
9
934
Carlos Baquero retweeted
Should you do a PhD? What would it give you? Here are some advantages to consider: 1. It’s a loophole in the universe where you’re encouraged to dedicate your time to something just because it’s interesting, even if it is totally niche and no one will ever make money out of it. 2. You’ll be surrounded by smart people. 3. It’s a gym for the brain - you’ll develop critical thinking and planning skills (extra important in the age of brain rot) 4. You might get to travel and see the world. … 786. Some jobs require having a PhD.
9
40
347
28,572
Carlos Baquero retweeted
So apparently it exists now: rstation.ai but they are not yet orchestrated as in Sakana's Agentic Scientist (pub.sakana.ai/agentic-scient…). Interesting times.
Remember SETI or Fold@Home? We need OpenSource@Home, where we use the tokens that would expire in everyone's subscriptions to contribute to the public good.
1
4
338
Carlos Baquero retweeted
For 25 years people speculated the NSA hid a backdoor in the NIST curve seeds. Turns out the P-256 seed was candidate #1842 from a random number generator started at 01:52:34 UTC on August 16, 1997. Claude figured it out in under half an hour. The scariest thing AI has done to ECDSA so far is make it look less suspicious.
15
58
403
29,445
Carlos Baquero retweeted
Europe’s best climate and where to find them. All great options: but South of Italy is probably the most underrated
180
128
1,427
440,133
Carlos Baquero retweeted
Realmente as causas do aumento do preço da habitação em Portugal são um mistério
Por que é que as casas estão tão caras? Expliquei os motivos pelos quais a habitação em Portugal está tão cara. Não há oferta suficiente para a procura. Portugal deixou de construir. Como tal, a solução é mais construção! 🇵🇹
36
115
683
74,704
NO! Tesla and @elonmusk DID NOT invent the flying car. The flying car is a Polish 🇵🇱 invention. In 2020, a 41-year-old driver in Rąbień hit a roundabout at high speed, used the embankment as a launch ramp, flew roughly 64 metres, snapped a tree in half, and landed in a church building. No prototype. No venture capital. No Silicon Valley presentation. Just a Suzuki Swift, a Polish roundabout, and absolute faith in aerodynamics. Polish engineering did it.
AjaTollaH GPW
66
186
1,510
47,538
Reality is more amazing than AI.
2
5
350
Carlos Baquero retweeted
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
160
343
2,648
90,342
Carlos Baquero retweeted
We have new Speakersssssss!
3
3
349
We might still have the monopoly of free will and awareness. But we are no longer the most intelligent minds in our known universe. Let that sink in.
Now on @sciam: 5 weeks ago, OpenAI shocked the world by solving the "Navier-Stokes" math problem. Then two mathematicians said it had copied their work. Now it seems the AI-generated result is something stranger than either side had imagined. scientificamerican.com/artic…
1
3
4
1,044