Big G still got moves.
Huge drop from Koray Kavukcuoglu and the DeepMind team.
Koray, Demis, and Sundar are not playing around this year.
The new kids on the block thought they had the lead, but Gemini 4 Argon just dropped—and it’s taking over the leaderboards:
1 on the Vals Index (economic impact across coding, legal, finance, & tax)
77.9% on DeepSWE v1.1 (SOTA in long-horizon software engineering)
1 MILLION Token Output Limit (massive headroom for multi-step reasoning without cutting off)
When the tokamak powers on, something magical happens in the confines of its reinforced belly.
Plasma born of heavy hydrogen blooms into brilliant pink, roiling and writhing around the doughnut-shaped chamber.
Then, tiny grains of lithium tumble into the maelstrom, sparkling like scarlet fairy dust before bursting into streaks of vivid green-yellow light.
Finding alternatives to precious metals in chemical manufacturing may be less like sourcing a substitute and more like tuning an instrument already in hand.
Yokohama National University scientists have found that an earth-abundant cobalt catalyst can selectively hydrogenate nitrogen-containing compounds when the balance between metallic cobalt and cobalt oxide is carefully tuned. Their findings could allow for more sustainable approaches to chemical manufacturing that do not depend on precious metals.
The study is published in the Journal of the American Chemical Society.
Prions are one of nature’s cruelest tricks.
Proteins are chains of amino acids that need to fold in very specific ways – a process that does not always work as intended. Your body is pretty good at correcting these errors – but every so often, a prion emerges.
This isn’t just a crooked, non-functional protein. It’s a gnarly shape that shares instructions for how it got that way, while resisting the body’s clean-up mechanisms. As these instructions spread through the brain, the result is rapid, progressive degeneration and, ultimately, death.
I’ve uploaded my new paper to PhilSci-Archive. Click the comments section for the Medium article (with pictures!).
This basic premise came to me seven years ago in a San Diego hot tub. The word “problem” in its (admittedly provocative) title can be interpreted as a fatal flaw or an obstacle that must be acknowledged before further progress can be made.
The piece was always meant to be a conceptually over-enriched Necker Cube, a playful excoriation of the unimaginative or less-than-rigorous treatments I’ve seen of the simulation hypothesis from critics and acolytes alike (though there are exceptions, like Roman Yampolskiy’s papers).
At 1,700 words, it could have been substantially longer. However, brevity is a virtue. Let the reader extrapolate (and mutilate) to their heart’s content; that’s the fate of any attempt at art or philosophy anyway.
I will use my yearly @highlight for this.
