1 Like
“Don’t Die Yet”
Believe me, I’m trying not to.
Unfortunately my age puts me right on the cusp of new life-extension breakthroughs.
9 Likes
He mentions his 86 year old mother a couple of times, and I get the impression he thinks cutting edge AI is advancing so quickly that there’s a possibility she could benefit. Near the end of the video, he’s asked what would be his first prompt to an AI created without limits to the resources it would take to build it. He said he would ask what intervention would extend an elderly person’s life for 5 more years. The 5 years fits with his earlier suggestion that life extension will advance in increments of a few years at a time, allowing a person to still be alive when the big thing gets here - same as Aubrey de Grey has said. He threw out 5 years, 10, 3 - hard to tell if he’s overoptimistic or realistic. He says he’s witnessing AI capacities growing at a blindingly fast rate, and he’s there to see it, so I don’t think he’s just dreaming.
3 Likes
I would ask what infrastructure and regulatory upgrades we need to best facilitate the studies needed to gain the dataset necessary to reach longevity escape velocity. If necessary, what public relations campaign would be needed to convince the world to invest in these upgrades.
We would need massive warehouses of AI controlled robotics running wet lab experiments on all model organisms with a library of all known molecules. You’d almost want to start from scratch with what we think we know and just have these robots working 24/7.
Once something works well it enters into the next round where it is combined with other things that work well. Things that alleviate some hallmarks of aging but have some negative effects are sought to be combined with something(s) that offset those negatives.
In addition to this effective pace of aging clocks and age clocks would enable even faster lifespan trials to occur not only in model organisms but also humans.
Once all known molecules have been run through this program we can then task AI to synthesize new molecules based on what it knows about the tested molecules.
In addition, we would want to begin the process of true in silico technology to simulate the cell and biology and have billions of theoretical trials going on at once.
3 Likes
I think it is relatively obvious that this is not true. Any identified pathways that have effective interventions will affect lifespan. There is no reason why the lifespan effect should be constant. You could have one set of interventions that have a big effect and another set that have a small effect.
4 Likes
I think this is all quite silly. I have a radically different idea - perhaps also silly.
My “insight” is that all this is barking up the wrong tree. Trying to find molecules to affect aging is like trying to find better training methods that will let a race horse run faster. Tremendous effort for marginal gains.
The idea: if AI is getting powerful enough to find new molecules, it’s a tool that can be used differently - instead of finding a faster horse, we need to invent a car. Take all those resources and go directly into something that will actually move the needle. Genetic engineering and redesign. Life extension of 4-5-10 times. The greeland shark lifespan is some 400 years, or roughly 4 times longer than human. But look at the naked molerat. A mouse lives some 3 years. The NMR lives 30+ or 10 times longer - it is also a small rodent just like the mouse is. What is different is a different genetic makeup. What if we created in humans a different genetic makeup with a 1000 lifespan? There is nothing inherently impossible in this, no real biological reason why you couldn’t do this. Today we have tools like CRISPR that allow pretty precise rearrangements of the DNA. What we need is a mapping out of a design that allows negligible aging (as in some animals) and a method to introduce it gradually in currently living people. Naturally, it would be easier to do it from scratch before birth, but hey let’s get ambitious and try to do it in the already elderly - it would really suck to be the last generation of the 100 year lifespan humans.
My thesis is that the complexity of finding molecules that extend life by a few years is not appreciably smaller than going directly into genetic redesign. Use AI and the resources that go into novel molecules to get us a Naked New Man of a 1000 year lifespan. It might actually be conceptually simpler than thousands of wet labs hunting for molecular needles in a haystack which even if found will give us tiny results. Go for the gold, or go home. I personally have little faith in the “finding molecules” approach, and take drugs (including rapamycin) simply because there is nothing else. But I’d much prefer to exchange this broken down racehorse for a brand new race car.
8 Likes
It’s a long video. He does talk about genetic engineering and quite enthusiastically. He discusses molecules and diseases and lifestyle as well. The theme is AI and what it can and will do, not any one type of intervention.
1 Like
Not much to learn from this interview for me, except maybe as an advertisement for the latest OpenAI model. But maybe it’s because I hear this kind AI hype often in my industry. Funny he kept mentioning self-driving cars as a success/target, at best they have had mixed results so far.
This reminds me of when for awhile Attia would ask his guests what the impact of AI was on their work, and the most of the answers were just as generic as you could get.
3 Likes
Jackpot! When I posted the video, I just thought of it as an optimistic take on how soon we’ll get to Rejuvenation Day, but turns out it’s more than that. It’s controversial! And I posted it! I’ll treasure this day forever.
1 Like
I don’t see how your idea of genetic redesign applied to currently living people could be implemented any time soon. On broad population? On currently living elderly? It’s a great futuristic idea but at this point we can only fantasize about it. I’m pessimistic about it.
This was a great podcast to listen to and I sure hope that Dr. Unutmaz is correct. His optimism is evident throughout!
1 Like
And @LaraPo
Full blown genetic engineering could be done in a comprehensive and good also for older people if combined with/implemented through replacement, see eg
2 Likes
