Login

Sign In

Enter your details below.

Protein Folding: A Stellar Accomplishment for DeepMind AI

In recent Daily Prophecies, I have covered thrilling news from my COSM conference.

In a portent of Google’s afterlife, both Craig Mundie — recent Microsoft CTO — and Ray Kurzweil of Google reported that Google’s DeepMind laboratory had used a variety of expert disciplines in an artificial intelligence program called AlphaFold AI. The AI was created to master the baffling geometrical challenges of protein-folding.

Opening a new era in biotech and individualized medicine, this announcement is vastly more impressive than DeepMind’s acclaimed performance in games, beating unaided human champions in chess, Go, and various video games.

In Go and other deterministic games, gigahertz iteration speeds alone enable exploration of the bounded possibility space of symbol systems. In a game, the symbols and the objects are the same. With the “map” the same as the territory, games obviate the problem of interpreting the symbols and integrating them with the real world.

The success with protein folding, by contrast, represents a huge real-world accomplishment for DeepMind’s AI team, which refrained from trans-human hype and ascribed its success to multidisciplinary expertise.

Using AI machine learning techniques to predict the folds from DNA codes alone, the DeepMind group won the bi-annual Critical Assessment of Structure Prediction (CASP) competition, defeating 97 rivals. It succeeded in calculating the three-dimensional shapes of 25 out of 43 proteins, compared to the second-place finisher, which solved only three.

In the past, researchers have used various ad hoc methods, from nuclear magnetic imaging to x-rays and cryogenics, to identify the configurations and contrive the folds. With virtually infinite possibilities, progress was slow. Mostly using trial and error, these efforts entailed millions of dollars and years of work for each folded protein.

Protein folding stood as one of the grand challenges of science.

This breakthrough stands as one of the all-time greatest feats of computer science, and computer scientists. It indicates Google’s purchase of DeepMind for $500 million in 2014 in a signature move by CEO Larry Page.

As biotech is transformed from chiefly a biochemical problem to an information science, Google now has to decide whether to become a biotech company. Our biotech panel at COSM under Jay Richards of Discovery Institute brought together Matt Scholz of Immusoft and Oisin, Lindy Fishburne of Peter Thiel’s Breakout Labs and Babak Parviz of Amazon. Discussing the Information Theory of Biotech there was not a biologist insight.

Social Networks: Friend or Foe?

Moving from the sublime to the political, I should confess here that I failed at COSM to alert my readers to the terrifying national security threat from TikTok.

I’ve talked briefly about this “threat” in previous prophecies.

Detected by aspiring Paul Reveres from both sides of the Senatorial aisle, the Chinese video app from ByteDance is moving menacingly toward an initial public offering.

The threat? It appears to be streams of saltant dancing girls shaking their personal databases while recording the IP addresses and particular twerking preferences of users. Checking it out for my readers, I admit I got distracted. Amid my excitement, I may have missed something “existential” in all the bumps and grinds.

Perhaps the Senators do not want their special finds and favorites to be divulged to other connoisseurs, such as voting constituents. Or perhaps they believe that programs that collect information on the tastes of their users represent a national security peril.

That would help explain their similar panic over Google, Facebook, Amazon and other US companies that adapt their offers and ads to the revealed interests of their customers. It is this feature that makes social networks possible and profitable and may in time enable them to offer ads that you want to see.

Today nearly all the ads, particularly in smartphones where they use some 30% of the bandwidth, are in fact minuses. They are clicked on chiefly in error. As smartphones become the dominant computer screens, Google and the others have to solve this problem. TikTok’s dancing girls offer an age-old remedy that even a Senator can understand.

Comparing the US version of TikTok to the Chinese version, I can report that the Chinese girls were far less salacious than their US rivals, including Chinese in America. Here is a competition we are winning.

At COSM, however, we gently suggested that the privacy concerns of people publicly posting on social networks did not reach the level of a national grand challenge.

We also pointed out that the chief threat to the Internet is the quest for “privacy” interpreted as anonymity. With anonymous users of Internet Protocol addresses, the Internet cannot be secure, regardless of who provides the apps and the equipment.

At COSM, we banished anonymity in our COSM architecture introduced by voice-over-IP pioneer Daniel Berninger and internet visionary Giuseppe Gori. They take IP addresses ultimately controlled by national politicians and would replace them with PKI (public key infrastructure) apparatus.

The addresses will be public keys that are global and selected from a mathematically infinite lexicon as unrepeatable as protein folds. Transactional access is through private keys stored and processed in Trusted Platform Module (TPM) vault chips and not even necessarily known to their users.

Already on billions of devices, TPMs and their like are processors and nonvolatile memories physically isolated in the silicon and rooted to a particular identity on an immutable time-stamped blockchain. The user is not anonymous but unimpeachably linked to the chain by biometric authentication.

Takeaway: With real identities, the entire blockchain system becomes far more efficient, scalable, stable and safe. Comparable to DeepMind’s calculation of protein folds, the creation of such an un-hackable new internet architecture is a cosmic advance that promises to banish much of the paranoia about foreign apps and routers and IP theft that is stultifying governments around the globe.

Oh! And one last thing…

I just released the October issue of The George Gilder Report.

The issue features a pure-play that’s currently redefining digital storage.

This company is going to continue to lead the pack in customer loyalty, subscription renewals, and strategic innovation — benefiting more than most from the market’s impressive growth.

Luckily there’s still time to invest.

 

Regards,

George Gilder
Editor, Gilder’s Daily Prophecy

You May Also Be Interested In:

George Gilder

George Gilder is the most knowledgeable man in America when it comes to the future of technology — and its impact on our lives.

He’s an established investor, writer, and economist with an uncanny ability to foresee how new breakthroughs will play out, years in advance.

And he’s certainly no stranger to the financial newsletter...

View More By George Gilder