@mweinbach So succinctly put. I used to quickly parse over Gemini's outputs because they felt needlessly winding, but with Neon, every word feels intentional. Quite an amazing model, truly
Oh my god: Anthropic mathematician Levent Alpöge (@__alpoge__) says he has proven the Riemann Hypothesis, open for 167 years.
"I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis."
He says the method also covers the Generalized Riemann Hypothesis. The proof is out as Lean code (a proof language where a computer checks every step), so experts should be able to check it quickly, Scientific American reports.
Same Alpöge who was behind this year's Jacobian conjecture result and the Navier-Stokes work with Tristan Buckmaster.
I assume we will get an official statement very soon from him!!
Systems biology is the study of complex biological components (genes, proteins and enzymes, etc.) using holistic and quantitative methods; in my case above, I used Argon to better understand bacterial chemotaxis and the landmark 1997 Barkai-Leibler proposal which made its mathematical model robust. Hope this helps!
@exergiovanni@justinskycak But isn't that the whole argument, though? Some people will simply be more exceptional than others owing to the capabilities they inherently possess. And @justinskycak isn't saying that it's bad, just that it's true.
@sneycampos@OfficialLoganK@GoogleAIStudio Just used Argon earlier to ask questions about systems biology. Can't believe its last words to me were about attractant concentration.
@treetowntree@_a9lim So cool! Being able to cleanly ablate or swap specific concepts (like refusal patterns) without training seems handy for safety experiments. How does this method affect general capabilities/reasoning?
It is quite sad how Terence Tao practically made the same point a while back but got memed on by pro-AI folks because they interpreted his argument as "AI bad bcoz AI is taking muh job"
AI is a revolutionary tool, yes, but it is also the most efficient at increasing the cognitive debt of our supposed critical thinkers; it can lift people up to become "researchers", just not always a good one.
I've been talking to a bunch of PhD students and researchers recently, and I've qualitatively noticed two growing (but concerning) trends with AI:
1. Researchers seem to be reading less. Some even reported that they read <10 papers a year now with AI because they can simply either 1) retrieve or 2) summarize papers. BUT deep reading is how we engage with underlying intellectual arguments, develop research taste, spot methodological flaws, and form original ideas. It's arguably the most fun part of a PhD imo
2. Researchers seem to be vibe-coding and paper-maxxing. AI dramatically cheapens execution, such as implementing experiments, running analyses, and producing manuscripts. Some even "let [Claude Code] rip" without checking the code or validating the results somehow. BUT it also incentivizes quantity over depth, rigor, and meaningful contributions.
Overall, the cost of producing research is falling much faster than the cost of developing genuine intellectual understanding. Rather than simply producing more papers, I think we should engage deeply with the literature, ask questions that truly matter, and produce work that meaningfully advances our collective understanding.