7 Aug 2026, 21:38 UTC243 views2 reactionsread 8 August 2026 Nature reports that AI systems are already finding real errors in scientific papers and even decades-old reference databases, including incorrect boiling points, typos, calculation/figure errors and failed reproductions.
Important caveat: some of the cited evaluations used very outdated models, such as Claude 3.7 and GPT-4.1, or non-maximal reasoning settings. So the reported limits are better viewed as a lower boun…
👍1🥱1
7 Aug 2026, 17:44 UTC381 views10 reactionsread 8 August 2026 Video
Video, posted without a caption
😁3🔥2🤡2👍1👎1💯1
7 Aug 2026, 17:36 UTC860 views6 reactionsread 8 August 2026 Video
Video, posted without a caption
😁3🤣2🤡1
6 Aug 2026, 22:02 UTC≈1,190 views2 reactionsread 8 August 2026 Some links:
1. Scientists at the Arc Institute in Palo Alto have used AI to successfully create entirely new kinds of viable viruses that have never existed before in nature. The study was published today in Science. Reporting by the New York Times. https://www.nytimes.com/2026/08/06/science/ai-viruses-bacteria-arc.html [no paywall: https://archive.is/GMMiv / paper: https://www.science.org/doi/10.1126/science.aec265…
👍1🤡1
6 Aug 2026, 21:03 UTC433 views15 reactionsread 8 August 2026 Russia's useful idiots, such as Mearsheimer:
- Ukraine has lost 10 million soldiers.
Russia:
- Mercenaries made up of degenerates and criminals capture Rostov-on-Don without resistance after their warlord bemoaned gigantic losses. In a panic, excavators tore up highways towards Moscow.
- Gets literally invaded by Ukraine and needs North Korean soldiers to liberate its territory.
- In the fifth year of the 3-day …
😁9🥱5💩1
6 Aug 2026, 17:52 UTC453 views4 reactionsread 8 August 2026 Video
Video, posted without a caption
👍3🆒1
6 Aug 2026, 16:32 UTC482 views5 reactionsread 8 August 2026 Photo
Photo, posted without a caption
🥰5
6 Aug 2026, 13:52 UTC487 views3 reactionsread 8 August 2026 Tony Feng on the 10 OpenAI Astra proofs:
I am a mathematician but I avoided commenting on this because the problems are (far) outside my domain. Even if I sat down to read the technical details (which I have not), it would be hard to appreciate the context of prior work, etc. But I've gotten enough sense of things through colleagues to be confident about some aspects.
First of all, all the mathematics is fundamenta…
🤡3
6 Aug 2026, 12:41 UTC≈1,360 views18 reactionsread 8 August 2026 Video
Video, posted without a caption
👍8🫡8🤮2
6 Aug 2026, 09:01 UTC511 views2 reactionsread 8 August 2026 Photo
Collaborations and "social" interactions emerge naturally in AI agents: https://www.cybersecuritydive.com/news/openai-hugging-face-hack-ai-models-black-hat/827167/
Quote: “we should expect that threat actors will intentionally deploy, optimize, weaponize, and use offensive agent collectives in the manner that we have just described here.”
🤡2
5 Aug 2026, 21:18 UTC680 views10 reactionsread 8 August 2026 Photo
Prime Agent: A self-improving RLM harness for coding and long-running autonomous tasks.
On ARC-AGI-3, it scores 95.5%, surpassing the human-expert baseline, but the gain is not benchmark-specific.
Prime Agent combines three ideas:
1. Recursive Language Models-native programmatic tool calling.
2. Persistent multi-agent orchestration.
3. A self-improving Continual Harness.
Together, they let the model act on its ow…
👍4🤯4❤1🤡1
5 Aug 2026, 17:34 UTC643 views7 reactionsread 8 August 2026 Photo
Top scientists leave Google to found a new company, wiping roughly $175-180 billion off its market value: https://www.discoveryloop.com/
Collectively, we represent three of the most-cited researchers in artificial intelligence and two of the most-cited researchers in distributed systems.
Between us, we have pioneered massive scale computing and led the creation of critical infrastructure, products, and foundational…
❤4🤡2🌚1
Showing the 12 most recent of 30 posts we hold for @axisofordinary. View and reaction counts are the latest single reading for each post, not a live figure, and a recent post is still accumulating both. A view count marked ≈ was rounded by Telegram before we ever saw it — t.me prints views in full below 1,000 and to three significant figures above, so ≈1,200,000 means somewhere between 1,150,000 and 1,249,999. Unmarked counts are exact. Text is reproduced from the public post preview and truncated for length.