Central Control over AI... and everything else

Back home, hoping for a rest after nine weeks of relentless travel and speechifying.  But no, I cannot lay off this unpaid scrivening, here. Because, well, the issues and misunderstandings are just to rife. For example: I was recently in China for a corporate conference, to keynote a session on “Daily Life in the Future.” There, I got to sample different cultural and political perspectives in Shenzhen from Hong Kong, just across the border.

Around the same time, the Washington Post published an opinion piece by a Chinese academic that has caused a stir, by offering a cogent and thoughtful – and ultimately wrongheaded – argument that the sole solution to AI – related problems will be control by a paramount party-state.

We need to understand how these polemical rationalizations — e.g. “we are getting even for colonialism”  — aim to gird and rationalize a heightened level of intensity. Intensity that's not needed, in order to develop an advanced and competitive nation… but that will be necessary if your aim is to stir militancy, even war.

== The argument for central control over AI… and everything else ==

Feng Xiang, a professor of law at Tsinghua University, argues that “AI will spell the end of capitalism.” 

First, the standard Marxian cycle will return, with a vengeance. For lack of anti-monopoly or redistributive reform (like those enacted by our parents, under FDR, or our great grandparents, under the other Roosevelt), each business cycle will result in greater wealth disparities and a narrowing of the owner-controlling caste, leading to a conversion of vibrantly competitive markets back into history's standard, uncreative oligarchic pyramid. 

Naturally, Professor Feng’s proposed solution is also Marxist. Party-guided proletarian revolution.

Second, technological obsolescence of many types of employment will break the livelihoods of hundreds of millions, if not billions. No longer able to negotiate or bargain for the value of their labor, workers will be at the mercy of the Owner Caste. And yes, ditto. Feng’s prescription for a resolution is Sino-Marxist.

Finally, any AI that gains control over important systems with unsupervised intelligence may pose an existential risk to humanity. For this and other reasons, Professor Feng argues that research into artificial intelligence should be tightly controlled by a benevolent socialist state.

Why am I giving space over to a communist state-servant who promotes Marxist notions that I clearly disagree-with? Because it is well-worthwhile reading his appraisal of the looming problems. After which it is instructive to study his prescriptions! Because simplistic panaceas will doubtless appeal to billions, over the next couple of decades. Especially when our own lords seem determined to follow the Marxian pattern, by driving the middle class into penury.

You need to grasp the polemical intent underlying Professor Feng's missive. And to see how Feng's prescriptions do not follow, logically, from his well-described premises.

(Note: for a much more detailed look at these rationalizations for recreating the feudal pyramid, see: Jiang Shigong's monograph on ‘Philosophy and History: Interpreting the “Xi Jinping Era” through Xi’s Report to the Nineteenth National Congress of the CCP’.)

== Zooming in ==

Let’s dive into Feng Xiang’s own words:

“But China’s socialist market economy could provide a solution to this. If AI rationally allocates resources through big data analysis, and if robust feedback loops can supplant the imperfections of “the invisible hand” while fairly sharing the vast wealth it creates, a planned economy that actually works could at last be achievable.”

Hold back your visceral reaction. Yes, yes, this is a blatant attempt to justify the ingathering of overwhelming power into a permanent, narrow, self-chosen caste. It’s the same model that dominated 99% of all human societies. Each of those also invoked incantations to explain why a dominant caste should hold and monopolize all power.

Almost anyone raised in the Western Enlightenment – or by Hollywood film morality – will feel instant recognition and loathing. Indeed, the difference between a state controlled by capitalist owner-oligarchy and a pyramidal hierarchy controlled by communist party elite is (basically) only one of vocabulary and incantations, not structure or end result.

But as I said; hold on a minute. It’s one thing to recognize a servile charm-chant to justify central power. It's quite another thing to dismiss every aspect of Prof. Feng's argument. We’d be fools to do so.

“The more AI advances into a general-purpose technology that permeates every corner of life, the less sense it makes to allow it to remain in private hands that serve the interests of the few instead of the many. More than anything else, the inevitability of mass unemployment and the demand for universal welfare will drive the idea of socializing or nationalizing AI.”

Complaining about the rapacious, insatiable and socially irresponsible behavior of today’s capitalist corporations, he asserts:

“These companies have been able to get away with their social irresponsibility because the legal system and its loopholes in the West are geared to protect private property above all else. Of course, in China, we have big privately owned Internet companies like Alibaba and Tencent. But unlike in the West, they are monitored by the state and do not regard themselves as above or beyond social control.”

In other words, Professor Feng proclaims that state planning will be boosted in effectiveness by the very thing (AI) that would be lethal, under capitalism.

Feng Xiang continues: “Marx’s dictum, “From each according to their abilities, to each according to their needs,” needs an update for the 21st century: “From the inability of an AI economy to provide jobs and a living wage for all, to each according to their needs.”’

Why have I given lengthy space to this? Even urging you all to go read the original, for yourself? Because of the extreme (and interesting) cognitive dissonance of this essay, which accurately describes a trio of serious dilemmas that have us clearly on course to failure...

...only then Feng explicitly declares that the only answer is to return to a different failure mode! A governance method that proved universally calamitous across 6000 years.

== An aside: the “central planning wall” ==

Consider that state planners know there is a “wall of incompetence.” Such a wall tripped up the Soviets in the 1950s, when their command methods for building primary industry (steel, dams, railroads) that worked so well in the 1930s proved inept at secondary industry capable of producing a refrigerator that anyone wanted. 

The 1980s Japanese zaibatsus – organized by the Ministry of Trade -- were convinced that their planned mercantilism had overcome Soviet errors, by using capitalist tools… till hitting their wall: the tertiary economy.

There can be no arguing with the fantastic successes wrought by the current Chinese leadership’s use of planning combined with corporate structures and predatory mercantilism. Even if their debt bubble pops tomorrow, their achievements in advancing their nation have been epic. 

Yes, a few of us point out that the fundamental key to their success was an American innovation – generously counter-mercantilist trade patterns instituted by the American pax since 1946 -- that uplifted half the population of the globe. Without that unprecedented indulgence by the era's "central kingdom," the “Chinese Method” would have gone exactly nowhere.  Still, the engineers who occupy seats on the current Beijing politburo are smart guys, and they can be excused some hubris, believing they have a way around the next state planning “wall.”

The magic tool they have used for a decade as been pre-AI computer modeling. And they expect that to transform into the ultimate wall-breaker – true artificial intelligence -- which will supposedly make economic and tech models so realistic that central planning will outstrip every system based upon Adam Smith’s markets or Friedrich Hayek’s ‘distributed wisdom.’

Do you now understand better the quasi-religious faith that central planners vest in the positive traits of AI?  Meanwhile, they posit – along with Elon Musk and many worried westerners - that there can be negative effects of burgeoning AI, as well! Even calamitous ones. And hence, Feng asks: who better to prevent Robopacalypse than a central party state?

Yes, yes, he follows his cogent dilemma description with a self-serving, magical incantation for centralization, without a scintilla of evidence or reason. No pyramidal power hierarchy ever evaded for long the core human contradiction: that we are delusional beings. Only one antidote has ever been found for delusion and error – free and open criticism. And tiny ruling castes always, always crush criticism.

They will hire gifted theologians – like Professor Feng – to concoct catechisms in whatever state religion justifies paramount power. But no matter how potent their AI, the fundamental remains the same. Garbage in, garbage out.

GiGo. There will be a wall.

== Don’t be smug ==

Do not use my own glib incantation about GiGo to dismiss all of Feng Xiang’s arguments! He is absolutely right that:

- We must find ways to avoid a gathering of all power into our own style of pyramid. One that’s inarguably worse – in its final stage – than Confucian paternalism. Feudal lordship by an owner aristocracy.

- Looming technological unemployment does mean that some renegotiation of the social contract will be absolutely vital.  As in Roosevelt’s time, it will involve some redistribution of ownership, likely vesting all citizens in shares of the means of production. (A much better solution than universal welfare income: read Vonnegut’s Player Piano, and Farmer’s Riders of the Purple Wage.)  The Greatest Generation knew this, as did the American founders, who redistributed a quarter of all the land in Britain’s former colonies.

But an American style reform would still entail the widest possible distribution and separation of power and influence. Indeed, although there will be screams of “socialism!” such a redistribution would be diametrically opposite to both types of pyramids. Both feudal oligarchy and Chinese Party hierarchy.

And yes, AI could either help or impede, depending on one thing. Transparency.

== Feng is also right that AI threatens us with peril ==

Want to see how creepy the present situation is?

In a 2014 article, Prof. Shawn Bayern demonstrated that anyone can confer legal personhood on an autonomous computer algorithm by putting it in control of a limited liability corporation. (“Independently wealthy software.”)  Such entities now operate independently, accepting and transferring payments and hiring humans for offline services.

In a fascinating article, UCLA Professor Lynn Lopucki asserts that algorithmic entities are likely to prosper first in criminal activities or those that benefit most from operating in the dark.  

This comes as no surprise to readers of science fiction. Autonomous algorithms featured in the novels of John Brunner and Joe Haldeman, long before gaining attention in William Gibson’s “Neuromancer,” wherein the protagonist only at the end realizes his employer was a cryptic AI. And that is just one of countless ways that new AI methods can only be turned benign if they operate purely under light. 

My favorite solution – a universal, international treaty – would solve all this in just two sentences:

“If you think you own something – from a car to a home, to a corporation, or ship, or algorithm – you must say so, identifying it openly. No shell entity may layer more than two deep before revealing owners who are living humans or totally transparent foundations.”

Two sentences.

Some have called this a “welfare program for lawyers,” since the litigation that it unleashed would last a decade. So? The biggest immediate effect would be a tsunami of suddenly revealed and abandoned property – never declared by the owners who acquired it through crime -- perhaps enough to give all legitimate taxpayers a tax holiday, worldwide. 

If you are a legitimate and honest taxpayer, there is no measure proposed by anyone that would benefit you more. I defend this proposal elsewhere. But here I assert simply:

* If it were imposed and enforced worldwide. No other measure would do as much to restore fairness and enlightenment and progress to the world. *

== The Final Feng ==

You all have been champing at the bit to point out the last and most glaring error of Feng Xiang’s missive. To declare that AI entities will be rendered harmless if controlled by a paramount party elite, atop an all-powerful central state. The error is obvious:

1. So empowered, that core elite will be unquestionable. A Big Brother that no Orwell could ever have imagined.

2. What happens when the AI gets truly much smarter than its masters? Won’t this power structure be trivially convenient for that entity to simply and subtly invert?  A pyramid of already unquestionable power, easily taken over by the very entities it claimed to control.

Jesus. Is any other outcome even remotely possible?

Yes, I prefer the distributed-competitive model, breaking up all elites and powers into reciprocally-accountable lumps. A method that works only under fully pervasive light. But when light does shine into all crevices, it works better than any other system ever devised.  (Why do you think that all the world’s oligarchs seek – in a worldwide putsch – to strangle the method, forever?)

Read Feng Xiang’s missive, with all of this in mind. There is partial value… like many things in life… but also silliness. Use light to separate them.

== Coda ==

In sharp contrast.