Is DNA next?
Create humans?
Important new research, but this is the most important point by far:
“Whereas you might spend six months running experiments to get a few dozen structures in a given cell type, you can generate a thousand structures in a particular region with our model in 20 minutes on just one GPU,” Schuette says.”
This is the problem and the danger with coming advanced science and engineering AI. At first it sounds pretty nice: Lots of progress in a short time, but too much progress too fast is disruptive and fatal even.
The reason the AI companies want AGI is because they already understand that this vast progress and knowledge is almost within their grasp and it is literally driving them crazy. It is a real life version of the ring of power that is corrupting everyone who learns of it. Suddenly, and unexpectedly, we are in completely uncharted and very dangerous waters.
I believe President Trump is at least partly aware of the danger, but it is not clear yet which way we will go. Keep in mind that during the start of the Covid pandemic, Trump at first listened to and followed the advice of those experts around him who advised “2 weeks to flatten the curve.” Also, masks, “6 feet” and allowing Moderna, for one, to market a vaccine that was developed almost literally over a weekend or some such brief time period.
The problem is that the “experts” like Sam Altman, Peter Thiel, Marc Addreessen and Larry Ellison and others are all advising Trump to accelerate AGI development so we can “beat China.” They are tech people, not geopolitical strategists. They seem to have no idea what China will do if we push them into a corner they feel they have no non-military way out of.
It turns out that it is highly unlikely we can actually “beat China” in in the end in an AI war. To understand why this is so, you have to understand the magnitude of what it means to truly lose the AI race to another country. It is, strangely to many, a national independence death sentence.
Knowing this, they would would rather in the end take out our AI infrastructure even though we would certainly do the the same to them. A limited nuclear exchange would likely set back both of our AGI programs for years, but we would both remain free, independent nuclear powers. Minus 25 to 50 million people on each side. THIS IS A PRICE THEY CAN PAY MUCH EASIER THAN WE, AND MORE WILLINGLY. We would both survive, but not as AGI powers. This is why it is impossible to “beat China” in the AI race.