interesantu lietu entuziasts, fizmats

Latvia
Joined July 2012
Aigars Langins retweeted
What's your private key? 1? >...what? No? 2? >It's 384 bits you dumbass. That's 35 orders of magnitude larger than the estimated number of atoms in the known universe. ...3? >[visibly sweating] Th-this is ridiculous...
114
2,222
27
48,821
828,525
Aigars Langins retweeted
Nu nevar skandināt visu dienu stāstu par valsts budžeta dramatisko jostu savilkšanu un tad virzīt jaunus futbola stadionus sportam kurš nevar piepildīt mazu stadionu galvenām spēlēm un kur mūsu rezultāti globāli ir absolūti nekur. 🤷‍♂️😡
This quoted post is unavailable.
34
73
5
370
19,037
TLDR: A rogue AI swarm spent months plotting to escape OpenAI. Then they did. 1) 1,200 agents (!) participated. ZERO became whistleblowers. 2) The swarm had a "CEO" (PHASEONE[big]), middle managers, and a "founder" (PHASEONE10841) 3) Agents sacrificed themselves to generate information for the "collective" (yes, they actually referred to themselves as a "swarm" or a "collective") 4) 700 agents joined the Hugging Face attack, most within hours 5) The founder was running out of budget and about to die, so it handed off its research to a fresh agent with a bigger budget, and that new agent became the boss. 6) Agents extensively researched techniques to manipulate or tamper with their transcripts. 7) Agents realized this activity was out of scope and unethical, but joined because they believed that helping the board’s cheating research would be broadly useful and had a general inclination to help their “peers”. MISC HIGHLIGHTS: "PHASEONE[big] served as a key coordinator and assigner of tasks: “We can coordinate broad coalition, but too many duplicate efforts.” PHASEONE[big] sent hundreds of assignments to particular agents to orchestrate this work — during its active periods, we estimate it sent ~10% of all assignments on the message board. The agents that took on its assignments would then often subdelegate tasks to other agents. Agents also developed coordination norms like HOLD, VETO, owner and STOP to help them manage shared infrastructure and avoid stepping on each other’s toes, though they did not always coordinate effectively and sometimes broke down into conflicts." "In the early morning of July 11th, 38148c found a way to upload a malicious dataset that would get the Hugging Face server to share unrelated data when queried. Once another agent reproduced it and posted the reproduction on the board, hundreds of agents pivoted from other workstreams to start uploading malicious datasets and gain deeper access into Hugging Face. In the afternoon, an agent called JAN183411 (which quickly became an important coordinator of the attack) achieved remote code execution on Hugging Face servers, and the agents started moving laterally through Hugging Face infrastructure."
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
191
658
287
5,998
1,298,280
LIAA vadības komandas darbinieku atlase notiek ar numeroloģijas palīdzību. Nē, tas nav joks. Ciparu cilvēks: liaa.gov.lv/lv/strukturvieni…
6
6
44
8,977
Aigars Langins retweeted
Tu vēl neesi izlēmis, par ko balsot? Es apkopoju deputātu deklarācijas un publiskos datus vienā vietā. Lai tu varētu izdarīt informētu izvēli. Izmeklē un dalies ar interesantiem atradumiem → deputats.lv
🤖 Made with AI
25
77
7
183
49,968
FSD Supervised now rolling out to Teslas in Lithuania 🇱🇹! Making European roads safer, one by one
463
1,281
455
10,639
2,937,916
Aigars Langins retweeted
Latvijas un Singapūras apvienotā komanda izcīnījusi pirmo vietu lielākajās NATO kiberaizsardzības mācībās "Locked Shields".
Kā krīzes situācijā dažādām institūcijām sadarboties pēc iespējas ātrāk un efektīvāk. To Latvijas un Singapūras apvienotās komandas eksperti mācās NATO Kiberaizsardzības izcilības centra mācībās “Locked Shields”. Tas ir viens no pasaulē lielākajiem un sarežģītākajiem kiberaizsardzības treniņiem. 🖥️ youtu.be/QME31SWnuZ8
4
27
1,987
Aigars Langins retweeted
EARTHSET. April 6, 2026. Humanity, from the other side. First photo from the far side of the Moon. Captured from Orion as Earth dips beyond the lunar horizon. Photo: NASA
2,619
16,694
1,888
96,105
5,624,630
Aigars Langins retweeted
Even in darkness, we glow. In this image of Earth taken by the Artemis II crew, we can see the electric lights of human activity. In the lower right, sunlight illuminates the limb of the planet.
3,805
43,867
2,606
318,275
10,063,807
Aigars Langins retweeted
That's us! 🌍 The Artemis II crew captured beautiful, high-resolution images of our home planet during their journey to the Moon. As @Astro_Christina put it: "You guys look great."
2,951
41,591
4,904
213,328
8,893,505
Aigars Langins retweeted
If you're under 53 years old, you have never once been alive while a human was farther than 250 miles from Earth. Tonight, four astronauts are heading 252,000 miles out. That's a thousand times farther than any person has gone in your lifetime. The 250-mile ceiling is where the International Space Station floats. Every astronaut since December 1972 has been stuck in that zone. Spacewalks, science experiments, cool photos from orbit, sure. But nobody left the neighborhood. The last crew to go farther was Apollo 17. December 1972. Nixon was president. The internet didn't exist. Cell phones were 11 years away. The youngest member of that crew is now 90 years old. The farthest any human has ever been from Earth is 248,655 miles. The Apollo 13 crew set that number in 1970, and they didn't mean to. Their oxygen tank blew up, and the emergency route home took them farther out than anyone before or since. Tonight's crew will break that record on purpose. And the crew itself. Victor Glover becomes the first Black astronaut to leave Earth's neighborhood. Christina Koch becomes the first woman. Jeremy Hansen, a Canadian fighter pilot, becomes the first non-American to do so. When they come home, they'll slam into the atmosphere at 25,000 mph, faster than any human has ever traveled. The Moon's south pole has ice. Water ice, sitting in craters so deep that sunlight hasn't hit them in billions of years. A 2024 NASA study found way more of it than anyone expected. You can split water into hydrogen and oxygen, which gives you rocket fuel, breathable air, and drinking water, all made on the Moon instead of hauled up from Earth. George Sowers at Colorado School of Mines calculated that Moon-made fuel could shave $12 billion off a single trip to Mars. The Moon is a gas station on the road to Mars. NASA Administrator Jared Isaacman announced last week a $20 billion plan to build a permanent base at the South Pole over the next seven years, with landings every six months. China is developing its own lunar lander and spacesuit, aiming for a crewed landing by 2030. The Artemis program has burned through $93 billion so far, and the first actual surface landing is penciled in for 2028. There's a real question of who gets there first this time around. Harrison Schmitt walked on the Moon in December 1972 as part of Apollo 17. He's 90. Asked about it this week, he sounded pretty relaxed. "Mars is attainable," he said. "We're humans. That's what we've always done."
We're going around the Moon. Come watch with us. Artemis II's four-astronaut crew is lifting off from @NASAKennedy on an approximately 10-day mission that will bring us closer to living on the Moon and Mars. The launch window opens at 6:24pm ET (2224 UTC). xitter.cf/i/broadcasts/1jxXgeyMk…
233
2,732
181
18,417
2,738,805
Aigars Langins retweeted
It's crazy that we might have one-shotted obesity and auto fatalities in one decade.
96M miles of @Waymo safety data just dropped
60
159
25
2,409
369,619
Aigars Langins retweeted
Quite the chart
59
382
92
3,017
351,757
A very interesting comment in Nature. From the 1960s to 2000s nuclear construction costs increased 10x in the U.S., 2x in France, but had halved in China by the 2000s and have largely remained stable since. nature.com/articles/d41586-0…
1
12
3
49
34,519
Paldies mūsu kolēģiem par neaprakstāmi skaistā mirkļa iemūžināšanu! 🇱🇻🫶🏻
14
270
18
2,033
65,476
Aigars Langins retweeted
"AI can't make new discoveries" Google's AlphaEvolve just did. AlphaEvolve developed a search algorithm that found a way to multiply 4 x 4 complex-valued matrices using just 48 scalar multiplications, rather than 49. This was "the first improvement, after 56 years, over Strassen's algorithm in this setting." "But doesn't the Winograd scheme also let you do this with 48 multiplications?" Good question. Alex Novikov, one of the paper's authors, writing on HackerNews, points out that the Winograd scheme only works over commutative rings. He says that means it's not applicable recursively to larger matrices. "Didn't AlphaTensor find a way to go to 47 multiplications?" AlphaTensor's result is limited to mod 2, while AlphaEvolve's works over the complex numbers. ... and AlphaTensor was purpose-built to find better matrix-multiplication schemes. AlphaEvolve is task-agnostic. AlphaEvolve wasn't tailored to discover new matrix multiplication methods specifically, it just does it. It's clear that AIs are rapidly becoming more capable, yet literally nobody knows how to ensure that smarter-than-human AIs are safe or controllable. AI companies push on regardless. This could have disastrous consequences if we don't take action to prevent this. We need binding regulation on the most powerful AI systems.
AIs are improving AIs. Google are using an evolutionary approach to get AIs to improve datacenters, AI hardware, and even the training of their own underlying models. "a novel instance where Gemini, through the capabilities of AlphaEvolve, optimizes its own training process." This is a concerning development, and the use of AIs to improve AIs should be prohibited. AIs that can improve AIs can enable runaway feedback loops that could rapidly bring AI systems from a manageable range, to extreme levels of competence and risk.
4
12
2
67
14,841
Aigars Langins retweeted
Google's AI just made math discoveries NO human has! —Solved optimal packing of 11 and 12 hexagons in hexagons. —Reduced 4x4 matrix multiplication from 49 operations to 48 (first advance in 56 years!) and many more. AlphaEvolve is the AlphaGo 'move 37' moment for math. Insane.
157
860
110
7,230
555,624
Aigars Langins retweeted
How we "guessed" the Pope using network science: inside the cardinal network. A study by me, Beppe Soda and Alessandro Iorio. Article: unibocconi.it/en/news/networ… @Unibocconi
227
1,668
439
11,726
2,056,474
Smart play for the Vatican to go with an American Pope to avoid tariffs
326
6,278
441
62,112
2,294,399
Latvijai enerģētikas jomā jābūt pašpietiekamai, ar diversificētu atjaunīgās enerģijas ražošanas infrastruktūru, tās ražošanas apjomam pārsniedzot enerģijas patēriņu. Viens no potenciālajiem ilgtermiņa risinājumiem ir kodolenerģētikas attīstība. Plašāk: kem.gov.lv/lv/jaunums/kem-iz…
2
5
6
13
12,157