The research behind this is wild. A baby owl can sit and starve to death right next to a pile of food. Put a stuffed owl next to it, like in the video, and suddenly it'll eat. An Austrian zoologist, Konrad Lorenz, won the 1973 Nobel Prize for figuring out why.
He showed that young birds aren't born knowing who their mom is. In the first few days of life, their brain takes a kind of mental photograph. Whatever they see moving around gets locked in as "parent." After that, only that figure can switch on their feeding instinct. He called it imprinting.
Owls have it worse than most birds. They're born blind, naked, and totally helpless. A baby barn owl needs feeding every two to three hours for weeks. It can't even keep itself warm until its feathers come in. And right around the time its eyes finally open, between days 15 and 20, its brain locks onto whoever's been taking care of it. Miss that window with the wrong face nearby, and the owl is wired wrong for life.
Even the begging is automatic. In the 1950s, a Dutch scientist named Niko Tinbergen ran experiments with baby seagulls. He found the chicks were pecking at a specific shape. A long thin thing with a colored spot was enough to trigger the full begging routine, even when it was just a painted wooden stick. Take the stick away and the whole sequence shuts down. The chick can be staring straight at food, but if there's no parent-shaped trigger, its body doesn't know how to swallow.
There's a tiny patch in the bird brain that runs this whole show. It's the same part that learns and stores faces. Researchers at Cambridge and labs in Japan have mapped it down to the chemistry. They've even found a hormone that, if you inject it in the right spot, can re-open the imprinting window after it closes.
That dummy owl in the video carries 40 years of conservation work behind it. In 1982 there were only 22 California condors left in the entire world. The San Diego Zoo started feeding hatchlings with hand puppets shaped like adult condors, hiding the human handler behind a curtain. The condor population is now 607. The Bronx Zoo did the same thing last spring with a baby king vulture. The Barn Owl Trust in the UK feeds orphaned owls through owl puppets while wearing camouflage hoods, because an owl raised by humans can never be released back into the wild. It'll fly toward people, beg from them, and starve.
The dummy is the only signal the chick's brain still accepts as "mom." Evolution carved a very specific lock into its brain, and only the right shape fits.
Rasmea Odeh became an American citizen and leader in her community, until zios targeted her, with the full might of the US government under their thumbs, to strip her of US citizenship.
Israel had tortured her mercilessly to confess to something she hadn't done. But she refused, withstanding unimaginable torture, until they brought her father into a room where she had been striped naked and they tried to force him to rape his own daughter.
That is who Israelis are.
That is what this abomination of a "state" does. Epstein was merely part of their global tentacles. We should spare these demons no terrible adjective, because not words can capture the depth of their evil.
Stage 3 rectal cancer in a 26 year old. This is what I've been warning about. The incidence of colorectal cancer in young adults has nearly doubled since the early 1990s. We're seeing Stage III and IV in people under 40 with increasing aggressiveness.
But the real story here is the treatment. This is dostarlimab, a PD-1 checkpoint inhibitor. Immunotherapy alone. No surgery. No chemo. No radiation. 100% clinical complete response rate in mismatch repair deficient rectal cancers. The immune system did what it was designed to do.
We've known since PD-1 inhibitors were approved in 2014 that this was the direction cancer treatment needed to go. The paradigm is shifting. Empower the immune system instead of destroying it.
Vercel will be officially sponsoring https://t.co/QF7eOed81b. That's a given. We as a community and industry owe @adamwathan and team a lot. Tailwind is foundational web infrastructure at this point (it fixed CSS 😉). I've also reached out to Adam to explore how we can make this a longer-term commitment.
I never thought Somalis from North Eastern would ever call out their leaders for impoverishing them while buying buildings in Nairobi. This is unprecedented.
Implemented Olmo 3 from scratch (in a standalone notebook) this weekend!
If you are a coder, probably the best way to read the architecture details at a glance: https://t.co/wF8PkoDuBe
Introducing Tahoe-x1 (Tx1) by @tahoe_ai. A 3-billion-parameter, single-cell foundation model that learns unified representations of genes, cells, and drugs, achieving state-of-the-art performance across cancer-relevant cell biology benchmarks, open-sourced on @huggingface. 🧵
# NewDataset for VLMs
After the release of VisualWebInstruct, we kept pushing its quality and adopting different strategies to make it as accurate as possible.
Today, we are releasing a verified version of VisualWebInstruct under https://t.co/ILfP6zmTlw.
It has around 100K verified STEM VQA triples. We got a huge improvement on top of Mimo-VL. Please give it a try if you are training VLMs.
Great work by @YimingJia486328!
BREAKING: Google's quantum chip computed a 150-year problem in 2 hours—and proved it.
Nature confirmed today: Willow achieved verifiable quantum advantage, meaning physics validates its answers. At 13,000× classical speed with 99.9% fidelity across 105 qubits, it's not just calculating faster—it's perceiving molecular structures that were mathematically invisible until this morning.
Announcing our beta launch: X API pay-per-use model.
We are expanding a closed beta to both new & power users who want to ship amazing apps on X.
All selected users will receive a $500 voucher to build with the X API. 🤑💻🚀
What if scaling the context windows of frontier LLMs is much easier than it sounds?
We’re excited to share our work on Recursive Language Models (RLMs). A new inference strategy where LLMs can decompose and recursively interact with input prompts of seemingly unbounded length, as a REPL environment.
On the OOLONG benchmark, RLMs with GPT-5-mini outperforms GPT-5 by over 110% gains (more than double!) on 132k-token sequences and is cheaper to query on average.
On the BrowseComp-Plus benchmark, RLMs with GPT-5 can take in 10M+ tokens as their “prompt” and answer highly compositional queries without degradation and even better than explicit indexing/retrieval.
We link our blogpost, (still very early!) experiments, and discussion below.
Announcing ClickGems, a free analytics platform for RubyGems!
https://t.co/dpeFM8BJhI
200K+ gems • 200B+ downloads 📊
Thanks to @rubycentralorg for making the data available!
Now with @metabase shareable charts on ClickGems & ClickPy — embed usage analytics anywhere.
I want u to hear these 2 doctors from Australia sending their last words, fully aware they could be killed any second.
What they’re describing is horror that defies belief.
70% of their patients are kids…