The Primary Focus of Innovation

George Gilder

Posted October 15, 2020

George Gilder

In both China and the United States, smart folk and fashionable finance regard artificial intelligence (AI) as the key to future dominance — both militarily and commercially.

If it doesn’t work as the experts expect, the results may be deeply disappointing and even devastating to your investments. Your portfolio, your computer, and your self-driving Tesla all might crash.

Fortunately, AI is not the only theme of transformation that offers a dramatic investment upside. There is also a coming “master algorithm.”

Three years ago, a leading AI theorist from the University of Washington, Pedro Domingos, wrote an influential book called The Master Algorithm that sums up the theory.

Algorithms, as he wrote, have inputs and outputs. In the human analogy, human senses collect inputs, and our brains and bodies produce outputs. Data enters and is processed into results.

Machine learning — the heart of artificial intelligence — reverses the direction as it trains the computer. Data on results and outputs enter from sensors, translators, and transducers, and the algorithm that relates them is found. This process performs for induction (from data to theory) what a Turing machine or logical computer performs for deduction (from theory to data).

Ultimately, Domingos believes all knowledge — past, present, and future — can be derived from data by a single universal learning algorithm, his “master algorithm.”

Meat Machine or Real Machine?

This general-purpose machine can run on combinations of NAND and NOR gates on silicon transistors on your computer. As defined by George Boole of Boolean fame and shown to work on simple transistor switches or relays by Claude Shannon in his historic MIT thesis.

Domingos’ book is a fascinating survey of the field and formulation of the problem. Explaining all the main schools of AI, from symbolic logic through neural “connectionism” (with "backpropagation" correcting errors), and genetic or evolutionary “fitness survival” algorithms, to predictive Bayesian “priors” and “posteriors,” he proposes adding it all up. A master algorithm might integrate the many special-purpose AIs into a singular general-purpose super mind.

The arrival of this machine, according to Ray Kurzweil, will represent a “singularity” in world history, as human beings fall inexorably behind thinking machines that can first replicate and then excel humans. Elon Musk’s company Neuralink is devoted to the proposition that human brains are essentially slow computers that general-purpose AI can program from the outside through a direct interface to the brain, a neuralink.

This is a powerful faith in the idea that humans are essentially “meat machines” that can ultimately be excelled by real machines. AI has nihilist philosophical roots that I describe in Life After Google and my new AI monograph for Discovery Institute as a “materialist superstition.”

Disabling AI is the paradox of its theoretical faith that theories are mere reflections of material and chemical relationships and thus are ultimately meaningless. Compounding the damage are ridiculous outbreaks of alarmism such as the late Stephen Hawking’s prediction that AI dooms the human race or Musk’s appeals for help from the government to keep AI under control or many calls for universal giveaways based on the idea that AI destroys jobs.

This philosophastering by nerds poses the danger of undermining technological progress by grandiose fearmongering. Whatever else it is, artificial intelligence is the next step in computer progress. If it is retarded or diverted by delusions the costs will be huge.

Therefore, prompted by Rich Karlgaard of Forbes, I was pleased to read Prediction Machines by three professors at Toronto’s Rotman School of Management and venture investors at their Creative Destruction Lab: Ajay Agrawal, Joshua Gans, and Avi Goldfarb. They understand that AI is just another advance in computing and like previous advances will hugely enhance human jobs, expand human opportunities, and increase human value.

An Important Lesson

They point out that AI is chiefly a prediction machine. Cheaper machine predictions will indeed reduce the value of substitutes such as human prediction. But cheaper predictions will increase the value of complements such as data selection and preparation, human judgment, and executive action. As AI advances, human minds will become more rather than less important and valuable.

The Twentieth Century saw a fundamental reordering of our understanding of the physical world from Newtonian determinism to quantum uncertainty. This led to the recognition that algorithms work best when structured probabilistically based on data.

The fatal reactionary error of the AI supremacists is to believe that a vast expansion of data and speed of computation will enable a return to determinism that ultimately obviates human intelligence and creativity.

But all economic and technical progress comes from human creativity, for which the signature is its unexpectedness or surprise. When a machine surprises us it is probably broken. Machines are determinist by nature. This flaw cannot be remedied by the current method of including random elements or averaging regressions that actually reduce real information.

The real master algorithm of the modern economy is not AI but information theory, which is based on the idea that all information is measurable by its degree of surprise. This master algorithm already lies behind all our achievements in computer science and communications. It even underlies all of real AI. It says that intrinsically predictable outcomes contain no real information. Unless they are breaking down, machines must be low entropy and predictable.

But Shannon’s information algorithms have a fundamental complement in his theory of cryptography from which they stemmed. The first use of the phrase “information theory” came not in Shannon’s postwar 1948 paper on communications and computing but in his previous — immediately classified — wartime paper on the “Mathematical Theory of Cryptography.”

To bear high entropy creative messages or content, a system must provide a low entropy ground state or carrier. Cryptography is designed to provide a reliable "ground state" for communication, preventing intruders from "hacking,” defrauding, or manipulating the essential premises.

Today’s Prophecy

The great challenge of information science in the next era is not to recover an impossible determinism through AI that eclipses humans. It is to integrate the low entropy carriers made possible with Shannon’s cryptography with the high entropy creations of entrepreneurs and innovators gauged by Shannon’s information theory.

The challenge is that the current internet architecture has no ground state. It is a gigantic copying machine. Over the next decade, the primary focus of innovation will be developing a new global internet and financial system with a ground state that cannot be hacked and manipulated by despots and central bankers.

Emerging over the coming weeks from a consortium of major companies will be a “COSM” architecture for the global internet. It will be based on the blockchain and public-private key identities and addresses that restore a ground state to the chaotic drift of the current global system. It will compete with the blockchain and digital currency innovations currently being launched in China.

The readers of my monthly investment letters will have unique access to the key players in this transformative project.

Stay tuned.

Regards,

George Gilder
Editor, Gilder's Daily Prophecy

The Price of Panic

Posted October 19, 2020

By George Gilder

Headlines everywhere are blaring — and lying — about “new cases!” Cases! Cases! These are not cases. These are positive tests, and the new ones are driven by the insane...

Chinese and Taiwanese Prosperity Should Be a Win-Win for America [Part 2]

Posted October 14, 2020

By George Gilder

If you missed part 1 of this Prophecy, click here to catch up.

As I was saying yesterday, this could be done relatively easily, in a way that would not...

Chinese and Taiwanese Prosperity Should Be a Win-Win for America [Part 1]

Posted October 13, 2020

By George Gilder

Recent US naval maneuvers in the South China Sea and the Taiwan straits raised the issue, “Should we risk war to keep China from subjugating Taiwan?”

Ever since President Nixon’s trip...

In Memory of My Mother

Posted October 12, 2020

By George Gilder

My mother died last week, at her home in her sleep, probably from aftereffects of her Boston Hospital visit that I described in an earlier prophecy. But I concede that...

Going to Extremes: The Reborn Potential of Technology Investment

Posted October 09, 2020

By George Gilder

Below, you’ll find an excerpt from The George Gilder Report.

If you’re not already subscribed, go here to get more information.


One of the themes of our...

COVID Update: It’s Not All Bad News

Posted October 08, 2020

By George Gilder

As usual, William Briggs has hit the mark when it comes to the reality we’re facing with the coronavirus.

Keep scrolling to read his thoughts…

It’s not all bad news, though there...