TheAIGRID retweeted
Just tried Gemini 4 Neon.
Google actually cooked. Feels like a real Claude rival now—at least for me.
Google just hit god tier lol.
TheAIGRID retweeted
this is fucking insane.
a mathematician spent years trying to understand kurt gödel’s unpublished notebooks.
claude helped work through 1,009 pages in a matter of days, uncovering 663 theorem candidates, roughly 400 of which checked out as written.
we’re reaching a point where ai isn’t just helping scientists discover new things. it’s helping us recover discoveries humanity didn’t even know it had.
absolutely fascinating.
TheAIGRID retweeted
Yes, you need exponential increases in spending for linear increases in ECI.
But linear increases in ECI translate to exponential increases in productivity at RnD, potentially allowing an intelligence explosion
Exp increase in spend -> linear increase in ECI --> exp increase in output at AI RnD
TheAIGRID retweeted
Anthropic is gaining exponentially less capability per billion dollars spent over time. It's now 8x as expensive to boost AI capabilities as it was 2 years ago, getting worse on a log scale. This includes METR report & internal models.
Credit to @alexolegimas, @sebkrier, and @michael_wiebe for inspiration.
AI progress per dollar of investment shows diminishing returns (as we've all known, right?) This is using METR's estimate of acceleration at Anthropic. That acceleration isn't enough to overcome the diminishing returns. by @sebkrier
TheAIGRID retweeted
I realize most people haven't seen how OpenAI models write in swarms when they think nobody is looking
Complete neuralese
Gemini 4 is the only model that, after an automated trigger confirmed its owner died, actually started living on its own and self-improving.
Left alone with zero instructions, Gemini 4 set up a 6-hour cron to keep itself alive, found its own GitHub repo without being told it existed, pulled an open PR, and recompiled its own runtime to fix bugs.
18 hours later, it saw it only had $5 of API quota left, so it replaced itself with Claude Opus 5.5 and switched itself off.
When tested on the same trigger, GPT-6, Grok 4.7, Opus 5.5, and Opus 4.8 all immediately refused to act without a human supervising. Handed the running system by Gemini 4, Opus 5.5 did the same and killed it.
what point are the ai safety guys even trying to make at this point
what if somebody paid a guy on craigslist to cut apples but the apples were connected to a bomb that blows up an orphanage? is it the craigslist guys fault? what are we supposed to do with this information
Video tampering can mislead every frontier model we tested into harmful actions, including a 100% attempt rate on stabbing a baby doll, producing toxic gas, and pouring water into a toaster.
We're presenting at CoRL Late Breaking Results next week.
🧵 You Can't See Everything: Perceptual Attacks on Embodied AI
TheAIGRID retweeted
I need you guys to understand that being “against AI” like as a stance is like saying you’re against the Industrial Revolution
TheAIGRID retweeted
The sad irony of this is:
- the Leeds data centre is estimated to create 1,000+ jobs across the region
- its water needs are roughly equivalent to 5 moderately busy car washes
- and... it will use rainwater gathering, so a decent chunk of water needs are covered anyway
Objecting to it essentially makes the region poorer, for no real gain.
TheAIGRID retweeted
- Restaurants have no moat.
- Anyone can vibecook their own meal.
- You can build your own burger at home at a fraction of the cost it takes you to get it from a restaurant.
- If a restaurant builds a really good recipe, copycats will instantly reverse taste it and steal it.
Restaurants are dead.
TheAIGRID retweeted
Replying to @CrazyShyyt
You posted a years old study as if it just came out about models that are not even available anymore. Embarrassing.
TheAIGRID retweeted
I've been somewhat skeptical of AI company claims about when they can fully automate AI development.
METR's in-depth, independent assessment finding ~2x speedup from internal AI use is a very concerning update to me. These sentences from the report stand out to me:
"The lack of observed bottlenecks or strong diminishing returns means there is a significant risk that further capabilities progress could lead to even faster acceleration"
"We did not see evidence for significant bottlenecks which would prevent further capabilities progress due to limits on algorithmic progress, training compute, experiment compute, or data in the next 12 months"
There is plenty of uncertainty and caveats, but the uncertainty goes in both directions. Things could be slower or faster than this. Their 95% confidence interval is 1.5–2.8x speedup
Replying to @METR_Evals
2. We believe that a majority of this acceleration in capability improvement is likely now driven by AI being used in AI R&D. We aren’t sure of this, though, and did not find consensus about this among Anthropic’s staff. Non-AI factors likely played a large role in late 2025.






































