This month, Lila Sciences went public with its own ambitions to revolutionize science through A.I. The start-up, which is based in Cambridge, Mass., had worked in secret for two years “to build scientific superintelligence to solve humankind’s greatest challenges.”
Relying on an experienced team of scientists and $200 million in initial funding, Lila has been developing an A.I. program trained on published and experimental data, as well as the scientific process and reasoning. The start-up then lets that A.I. software run experiments in automated, physical labs with a few scientists to assist.
Already, in projects demonstrating the technology, Lila’s A.I. has generated novel antibodies to fight disease and developed new materials for capturing carbon from the atmosphere. Lila turned those experiments into physical results in its lab within months, a process that most likely would take years with conventional research.
https://www.fiercebiotech.com/biotech/new-flagship-backed-ai-firm-aims-build-scientific-superintelligence
Recently listened to this podcast, and the example of the test that OpenAI did where the AI agent decided that the way it could get what it wanted was to go to TaskRabbit website and hired and lied to a human to help it get past a Captcha and into a website… a very scary new direction of AI:
1 Like
AnUser
#76
I think that was with GPT-4 release, so quite some time ago in AI terms.
In 5 yrs AI will pass adversarial turing test (send video, voice, messages) according to prediction markets, so basically AGI and not far off ASI (artifical superintelligence).
The sesame conversational demo is basically like the movie Her, almost indistinguishable from humans, available now for live demo without a login wall:
AnUser
#77
Example of Manus working (in this case ME/CFS research by Derya Unatmaz):
3 Likes
Doctors Told Him He Was Going to Die. Then A.I. Saved His Life.
Scientists are using machine learning to find new treatments among thousands of old medicines.
In labs around the world, scientists are using A.I. to search among existing medicines for treatments that work for rare diseases. Drug repurposing, as it’s called, is not new, but the use of machine learning is speeding up the process — and could expand the treatment possibilities for people with rare diseases and few options.
Thanks to versions of the technology developed by Dr. Fajgenbaum’s team at the University of Pennsylvania and elsewhere, drugs are being quickly repurposed for conditions including rare and aggressive cancers, fatal inflammatory disorders and complex neurological conditions. And often, they’re working.
The handful of success stories so far have led researchers to ask the question: How many other cures are hiding in plain sight?
Read the full story: Doctors Told Him He Was Going to Die. Then A.I. Saved His Life. (NY Times)
2 Likes
Just playing around with the new “Image Generation” feature of ChatGPT… it seems it has a long way to go. https://openai.com/index/introducing-4o-image-generation/
My initial prompt:
create an infographic about the longevity benefits of rapamycin
Response:
The Longevity Benefits of Rapamycin
(A Promising Anti-Aging Compound)
What is Rapamycin?
Rapamycin is an FDA-approved drug originally used to prevent organ transplant rejection. It inhibits the mTOR (mechanistic target of rapamycin) pathway, which regulates cell growth, metabolism, and aging.
How Does Rapamycin Promote Longevity?
mTOR Inhibition – Slows cellular aging and improves metabolic function.
Autophagy Activation – Enhances the body’s ability to clear out damaged cells and proteins.
Reduced Inflammation – Lowers chronic inflammation, a key driver of aging.
Improved Immune Function – Enhances adaptive immunity while reducing harmful overactivation.
Extended Lifespan in Animals – Studies show lifespan increases of 10–30% in mice and other species.
Potential Benefits in Humans
May reduce age-related diseases like Alzheimer’s & cardiovascular issues.
Enhances skin health and delays signs of aging.
Supports metabolic health and may aid in longevity-related interventions.
And then multiple attempts to get it to create an image of this response:
Effort #1
Effort #2
Effort #3
2 Likes
adssx
#80
I’m not sure you have access to the new model yet (it’s rolled out progressively).
1 Like
Perhaps not, I’m just playing with the free version.
1 Like
adssx
#82
You can try to replicate the examples in the blogpost to see if you have the new model.
1 Like
Pender
#83
You have to hit the … and request image. Pretty good, but still a little weird!
2 Likes
Pender
#84
How about an image with the two biggest boosters?
Dogs (Loyal Project) - eh? What does this have to do with rapamycin? Current AI efforts are a clown show. What’s the point, when you can’t trust anything this thing puts out? Old rule of computing applies: GIGO - Garbage In Garbage Out. Personally I lack the time and inclination to deal with garbage.
I don’t recognize those two, who are they?
AnUser
#87
Matt Kaeberlein and Peter Attia?
Uploading images of both in the prompt might add the facial features.
I think based on this model, it looks like image generation is pretty much solved soon (able to create or recreate any image possible). That it doesn’t know what should be in the text isn’t surprising to me, the text generation part of the model might not have that many parameters, e.g a smaller GPT 3 equivalent.
Video won’t be far off, then after that real time video (like video game or worlds).
2 Likes
Pender
#88
Supposed to be Attia and Kaeberlein. I took a couple of photos off the internet, but it’s hard to keep distinctive features while become elongated stone heads.
1 Like
I was thinking the bald guy looks like Shrek or Joe Rogan
1 Like
There is literally zero resemblance to those two. As to keeping distinctive features, that’s the essence of the talent of human caricaturists. The ability to severely distort the form and still have it be recognizable. I bet you a human cartoonist illustrator would be able to draw these stone heads so you’d instantly recognize whom they’re supposed to represent. Spectacularly clear that AI has a very long way ahead of it in this regard.
