Hey, it's lovely that they are at least trying to make overweight people "age slower". It's laudable. Of course, not really related to aging per se, at all, until proven otherwise. Not-killing-yourself (weight-normalization) just isn't a longevity drug and everyone who has seriously thought about the question should realize this**. From a longevity perspective it's odd to run the clocks for SELECT, or similar weight loss trials, and not for EVOKE - a population unselected for obesity. **you can do your post-hoc statistical magic to disambiguate weight-loss responders from those who show slower aging but it's still not a great trial design either way

Good read about one of the two recent disappointments in cardiology, the failure Lp(a) lowering, with the other being IL-6. Hopefully CETP won't be next with the obicetrapib trial reading out in the next 6 months.

The AI doomers were as right about capabilities as they were wrong about social and economic impacts of AI. We cannot trust them.

The God of Straight Lines is rarely wrong. What does this tell us about primordial prevention and the medical establishment?

All three GPT-6 models sit on the ZeroBench performance-cost pareto frontier pass@5 | pass^5 GPT-6 Astra (max) 51% | 35% GPT-6 Sol (max) 41% | 17% GPT-6 Luna (max) 20% | 7%

People have raised the question why certain accelerationists post with some level of glee about the current plight of mathematicians. While I can offer no complete explanation, I do think many people who are pro-AI want to celebrate their vindication after being relentlessly bullied by society over the last few years with a big "I told you so". Understandable even if perhaps not constructive. I just stumbled on a totally typical example of how AI-users and AI itself are viciously slandered as being below humans. Over the last years, I read thousands of posts like that. reddit.com/r/Filmmakers/comm… Imagine you believed that AI can make the world a better place, that it can reach human parity and beyond, yet people for years told you that AI is a toy, a stochastic parrot, worthless, disgusting, vile and then finally the receipts are coming in showing that you were right. A little Schadenfreude would be an understandable emotion, don't you think? Of course going forward both sides will have to keep their emotions in check, sit at the table and make a plan for a future that includes humanistic AI to better society.

If AMBAR used total plasma exchange, and geroscientists want to study TPE, and if - as it seems - neither has the money to run a large TPE study, why not join efforts? Repeat AMBAR in AD and measure clocks + function? Use AD + gero funding @BuckInstitute @msfuentealba @DrKiprov @EricVerdin @furmanbios

Verified U.S. clinicians can get free GPT-6 Astra (Pro) access today via ChatGPT for Clinicians. Sign up: chatgpt.com/plans/clinicians…

There are so many "bullshit" jobs in academia that can be 99% automated using AI. Jobs that should not exist to begin with. We need to spend more time on science and less time on bureaucracy.

If we peel away the unjustified vitriol against OpenAI from the Navier Stokes saga, there remains genuine uncertainty regarding training opt-outs. When OpenAI writes "While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models⁠." One obvious reading is "we have no idea whether you opted out from training or used the thumbs up/down button which overrides your opt out settings" I doubt that the frontier labs directly violate their own ToS. There are, however, rumours that the labs train on synthetic derivatives of customer data, despite opt-out, as a loophole. Seems unlikely, but not impossible. Probably a conflation of two things: I would assume they do train on synthetic customer data (which helps with de-identification) AND respect opt-out.

If AI has the power to destroy humanity, it also has the power to better everyone's life, solve climate change and cure aging. Harness that power!

I wanted to highlight this interview by Matt Kaeberlein discussing psilocybin not only because it is the kind of wild moonshot idea we need more of. It also contains a typical IACUC horror story. I heard and lived through so many of these stories that I now believe replacing IACUC with AI will be one of the biggest near term wins for society and science! youtube.com/watch?v=2VLZccpv…

Today's goal is to collect polyphenols like they are Pokemon.

A lot of public health research is masquerading as (healthy) longevity research. I love and respect public health, but we should not pretend it is something that it isn't. We need to spend more to understand and tackle the real biology of aging.

If you want to improve population health and avert demographic collapse, target aging.

The God of Straight Lines is rarely wrong.

Chess skill is remarkably stable with age, although it does decline. Perhaps a case of crystallized intelligence being more resistant to decline than other aspects of cognition. Just a quick and dirty analysis with obvious survivorship bias. LLMs are amazing for these back-of-the-envelope calculations!

Recently, I heard an interesting point about AI agents that can be paraphrased as "persistence masquerading as intelligence". But is it different from humans? Who solved Fermat's - was it a lone hero called Andrew Wiles or humanity through a century of incremental math progress?

As someone who has dedicated his whole live to curing difficult diseases, especially extending healthy long life, I have to concede that it will not happen without AGI. Everyone has to take their chances in life. This is our chance for a better future. Don't be afraid.

It appears that we need to teach long-running agents how to say "no, this task is not possible under reasonable parameters" and to ask for clarifications? Might help a bit. This has worked well for math, see e.g. BrokenArxiv progress.