Really interesting to have a comprehensive analysis of all the aspects that currently hinder faster experimentation with the biological side of that prediction. But I also think that the AI side will take way longer than most experts currently expect (they just project a shortterm exponential trend into eternity) and funnily AI needs to become even more biological to get to the levels needed to fulfill that prediction (or more honestly the "hope"). I've written a short piece about my view regarding this biological side of AI recently:
I'm imagining core facilities -- either one national center or many centers scattered around multiple schools, or perhaps even a company -- that acts as a CRO to synthesize and purify proteins and then test them in high-throughput. If there was a sufficient market demand, I think the cost could come way down. AdaptyvBio, for example, is making a sort of "chip" that can test and characterize hundreds of proteins in parallel.
"When scientists deleted the gene encoding a major Akt protein in mice, called Akt1, about 40 percent of the animals died as neonates while the other 60 percent survived and appeared entirely normal."
This is interesting - akt is central to the mtor pathway
"Mice lacking Akt1 display a 25% reduction in body mass, indicating that Akt1 is critical for transmitting growth-promoting signals, most likely via the IGF1 receptor. Mice lacking Akt1 are also resistant to cancer: They experience considerable delay in tumor growth initiated by the large T antigen or the Neu oncogene"
> One problem with this method, though, is that even a single cell is too thick for electrons to penetrate. But scientists today are using ion beams to shave off layers of the cell and thin it down to a diameter less than 300 nanometers, so electrons can pass all the way through. This technique "opens large windows into the cell’s interior," according to a 2013 review, allowing visualization of hidden structures that can’t be easily extracted or crystallized.
Yeah, I was wondering this the whole time, as you were talking about how we don't even know enough to create a virtual cell (which I actually didn't know, and hadn't thought about before).
Surely any self respecting ASI is going to come up with something like this - a comprehensive (if perhaps destructive) "total atom-by-atom scanning" process / machine that images complex 3d structures at maximal resolution.
You need massive storage and information-throughput processing ability, and I think the processing ability is the thing we lack and is why we haven't built things like this, but I think an ASI would kind of be defined by having a scalable information-throughput processing abiility that could scale up arbitrarily and remain integrated and useful.
Really interesting to have a comprehensive analysis of all the aspects that currently hinder faster experimentation with the biological side of that prediction. But I also think that the AI side will take way longer than most experts currently expect (they just project a shortterm exponential trend into eternity) and funnily AI needs to become even more biological to get to the levels needed to fulfill that prediction (or more honestly the "hope"). I've written a short piece about my view regarding this biological side of AI recently:
https://theafh.substack.com/p/brute-force-artificial-intelligence
> "industrialize and automate protein purification and testing in a more centralized way"
What do you mean by this? Better proteomics? In what way is purification lacking today?
I'm imagining core facilities -- either one national center or many centers scattered around multiple schools, or perhaps even a company -- that acts as a CRO to synthesize and purify proteins and then test them in high-throughput. If there was a sufficient market demand, I think the cost could come way down. AdaptyvBio, for example, is making a sort of "chip" that can test and characterize hundreds of proteins in parallel.
"When scientists deleted the gene encoding a major Akt protein in mice, called Akt1, about 40 percent of the animals died as neonates while the other 60 percent survived and appeared entirely normal."
This is interesting - akt is central to the mtor pathway
"Mice lacking Akt1 display a 25% reduction in body mass, indicating that Akt1 is critical for transmitting growth-promoting signals, most likely via the IGF1 receptor. Mice lacking Akt1 are also resistant to cancer: They experience considerable delay in tumor growth initiated by the large T antigen or the Neu oncogene"
> One problem with this method, though, is that even a single cell is too thick for electrons to penetrate. But scientists today are using ion beams to shave off layers of the cell and thin it down to a diameter less than 300 nanometers, so electrons can pass all the way through. This technique "opens large windows into the cell’s interior," according to a 2013 review, allowing visualization of hidden structures that can’t be easily extracted or crystallized.
Yeah, I was wondering this the whole time, as you were talking about how we don't even know enough to create a virtual cell (which I actually didn't know, and hadn't thought about before).
Surely any self respecting ASI is going to come up with something like this - a comprehensive (if perhaps destructive) "total atom-by-atom scanning" process / machine that images complex 3d structures at maximal resolution.
You need massive storage and information-throughput processing ability, and I think the processing ability is the thing we lack and is why we haven't built things like this, but I think an ASI would kind of be defined by having a scalable information-throughput processing abiility that could scale up arbitrarily and remain integrated and useful.