Thoughts on Joseph Weizenbaum's Computer Power and Human Reason

Between 2020 and 2025, approximately $1.87 trillion were spent on AI investment by global corporations according to the 2026 AI Index Report by Stanford University¹. This is an amount of money which is inconceivably large and nearly impossible for anyone to reason about. For example, the United States federal government spent $7.01 trillion in 2025², meaning recent trends in AI spending would make up over a quarter of the USA's budget (a country with ~345,000,000 people). Or, at a smaller scale, if we were to ascribe this figure to a single individual who spent $1 every second of their lives they would need to exist for nearly 725 lifetimes (or 57,970 years) to reach the moment they would run out of money³. With numbers like these it is difficult to imagine a future which is not, in some sense, saturated with so-called artificial intelligence. As the saying goes, AI is "too big to fail".

Most of the newfound prevelance of "AI" in the public discourse can be boiled down to a few bullet points:

• Google publishes Attention Is All You Need in 2017 which outlined the foundational transformer architecture⁴.
• Over the next few years, OpenAI and Google leverage transformers to develop GPT-1 through GPT-3 and BERT, respectively.
• OpenAI releases ChatGPT in late 2022 which triggers a massive funding race.
• By 2026, "AI technologies" have seeped into every last facet of society.

It is now impossible to have a conversation with anyone in America without it leading to a discussion about artificial intelligence⁵. These discussions range from the mundanities of how so-and-so's boss or such-and-such's company is now making them use AI tools, to the "shocking" revelations that AI "agents" have "gone rogue" and have begun "hacking their rivals"⁶. All of which contribute to the dull, thudding headache that one perpetually contracts from breathing air and reading the news. They point to one thing and one thing only: "This is how things are now! Enjoy, asshole!"

What's lost in many of these discussions, puffed up by overstatements and misunderstandings, is that many of these advanced computer programs are treated like they are anything but what they are: computer programs. At this point, it is clear that these programs have moved beyond the realm of parlour tricks, where they were relegated for some time, but what is unclear is what place, if at all, these programs have in society. Sure, GPT-6/Opus/Gemini/etc. may be capable of writing a technical report, proffering advice, assisting with image editing, etc., but should it?

Enter Joseph Weizenbaum. Weizenbaum, born in January 1923, was an emigre of Nazi Germany and an American immigrant who studied mathematics at Wayne State University and served in the U.S. Army Air Corps as a meterologist. After working at General Electric, where he developed programs used for automatic bank check processing, he was hired as a professor at MIT and was eventually awarded tenure. In 1976, six years after his tenure was finalized, he published his only major English work, Computer Power and Human Reason: From Judgement to Calculation wherein he outlined his thoughts on artificial intelligence and its limitations. Weizenbaum felt that "however intelligent machines may be made to be, there are some acts of thought that ought to be attempted only by humans."⁷ Before we can theorize about "intelligent machines", however, we must first understand unintelligent machines.

Karl Marx, the last great classical political economist of the 19th century, defines a "machine" as "a mechanism that, after being set in motion, performs with its tools the same operations as the worker formerly did with similar tools. Whether the motive power is derived from man, or in turn from a machine, makes no difference here."⁸ A machine, therefore, acts as a substitute for what the worker would have done without it. It functions as her replacement.

Why do we have machines at all? From whence do they come and why do they dominate modern life? Marx answers, "the development of labour into machinery is not an accidental moment of capital, but is rather the historical reshaping of the traditional, inherited means of labour into a form adequate to capital."⁹ A machine only needs to be operated, it does not need to be paid a wage, it does not need to be managed. The British economist David Ricardo wrote, "machinery and labour are in constant competition", adding that, "the former can frequently not be employed until [the cost of] labour rises."¹⁰ He continues, "the prices of commodities, too, are regulated by their cost of production. By employing improved machinery, the cost of production of commodities is reduced…"¹¹ In Marxian terms, "the machine is a means of producing surplus-value"¹². The more work a machine performs, the less work a human performs, and, by extension, the less value contained in the commodity produced. That means if a capitalist can beat others to adopting new technology, they can sell their commodities at parity with their competitors and pocket the extra profit. However, even when faced with the potential adoption of new technology, if the new machine does not increase the rate of profit "the capitalist, therefore, has no interest in introducing [it]. And since introducing it would make his old machinery simply worthless, when it has not yet worn out, transforming it into nothing more than scrap-iron, so that he would actually suffer a positive loss, he refrains from what would be, for him, a piece of utopian stupidity."¹³

The engine which affixes itself to the wheels of capital takes no interest in the toil of the value-producing class. What matters most is extracting as much value as possible. As John Stewart Mill writes, "It is questionable if all the mechanical inventions yet have lightened the days toil of any human being."¹⁴ This is simply not the job of machinery. In the event that the machine does so happen to lighten the load "the lightening of the labour becomes an instrument of torture, since the machine does not free the worker from the work, but rather deprives the work itself of all content."¹⁵ Where work becomes easier, it also becomes abject drudgery. The act of skillfully contorting leather to construct a shoe becomes the pushing of a button or the pulling of a lever.

The machine is always outside the worker—it is not a part of her, but rather she becomes part of it:

"The combination of this labour [social labour] appears just as subservient to and led by an alien will and an alien intelligence—having its animating unity elsewhere—as its material unity appears subordinate to the objective unity of the machinery, of fixed capital, which, as animated monster, objectifies the scientific idea, and is in fact the coordinator, does not in any way relate the individual worker as his instrument, but rather he himself exists as an animated individual punctuation mark; as its living isolated accessory."¹⁶

The worker is now indistinguishable from the machine. Her role in the labour process has been mechanized—she is little more than a tender to fixed capital¹⁷. In so doing, labour is appropriated by capital "in a coarsely sensuous form; capital absorbs labour into itself - 'as though its body were by love possessed.'"¹⁸

This absorbtion extends not just to manual labour, but to intellectual labour. Science, from the perspective of the capitalist, is the fuel for the great motor of history insofar as it propels great leaps in the development of the production process. As science is embodied in machines it then becomes the direct, physical antagonist to the worker. Science "appears only as means for the exploitation of labour, as means of appropriating surplus-labour, and hence confronts labour as a power belonging to capital."¹⁹ Once a discovery has been made, no additional capital is needed²⁰. Where the machine towers over the worker, so too does science. "Progress" comes at the worker's expense (and the capitalist's profit).

Despite impassioned pleas to the contrary²¹, a computer is nothing more than a machine. For Weizenbaum, "a computer is fundamentally a symbol manipulator"²². As many laymen know, the immediate "symbols" the computer "sees" come in the form of binary digits (or bits) which are stored electronically—i.e., as 0s and 1s. Various encoding algorithms dictate whether those bits represent a number, a sentence, or an image of the Mona Lisa. A modern computer is constructed of many different parts, both micro and macroscopic, which allow for the storing and processing of these bits to perform sets of desired computations. The lack of moving parts and their enshrinement in opaque cases lend a sort of "magical" quality to the innerworkings of the modern computing machine. Make no mistake, although a computer is a wonderous thing which contains an infinitude of sophisticated complexities, it is fundamentally not much different from the weaving loom. It is a box which receives input and provides output.

What makes the computer a historically significant machine is that the work it immediately began to replace was not physical work, but mental work. For example, one of the first electronic digital computers, the ENIAC (Electronic Numerical Integrator and Computer) was chiefly employed by the US Navy to complete artillery firing tables for gunships. Prior to its construction, many human "computers" would do exactly the same sorts of computations the ENIAC would do, only orders of magnitude slower with pencils instead of vacuum tubes. It would take the corporate world many years after the ENIAC's employment by the government to begin adopting computers en masse. Much of that had to do with their exhorbitant costs and extreme space requirements: the ENIAC cost around $8.9 million in today's money, took up about 1,800 square feet, and weighed roughly 25 metric tons.

As time soldiered on, more and more advancements were made in the realm of computer engineering. Transistors became the foundational technology for handling bits. New forms of long-term storage were developed—magnetic disks instead of mercury delay lines, and so on. The invention of integrated circuits made it so computers could be found in households instead of their traditional haunt of large office spaces. They became affordable, both for the individual and for capital (with many of these advancements coming from research funded by the United States federal government). The business world could now be bureaucratized to a degree that no one thought possible. Client information could be retrieved, cross referenced, and distilled in a fraction of the time it would take a statistics department to do so by hand. The intangible, mental work of analysis was absorbed into tangible, fixed capital.

Although computers were impressive in their technical complexity and powerful when set to the task of rote computations, they were unable to think—they could not act creatively. Many who styled themselves "computer scientists", before the term took hold amongst practitioners, sought to ascribe mind to matter and inject human-like thinking into such machines. Weizenbaum himself contributed to the literature with his papers How to Make a Computer Appear Intelligent and ELIZA — A Computer Program for the Study of Natural Language Communication between Man and Machine; the former, a description of a program which can play a restricted version of Go; the latter, a simple text processing program which mimics a Rogerian therapist. Throughout its not-so-long history, artificial intelligence research has focused on enabling computers to achieve human-like intelligence through the use of massive computation engines and mathematical modelling. In the terminology of the AI community, its nearest local maxima was reached in 2022 with the release of ChatGPT, a Large Language Model (LLM).

There is no one, agreed upon definition of what constitutes a LLM. The term is usually associated with "generative decoder-only (Transformer) models"²³, but it may be applied to models that do not use such technology or do not generate text. The most popular LLMs are often transformer based. "Transformers" refer to a neural network architecture with a focus on something called "attention"²⁴. Attention allows the model to quantify how strongly each word in a sequence relates to every other word, rather than processing the sequence one word at a time. Words or phrases with high attention to one another are treated by the model as more contextually related. Many transformers are layered sequentially beside one another, creating densely woven sheets of language processing units. The layers are then trained on vast corpuses of human text which set "likelihood" information called "weights" that help the program's predictive capability. All of these weights combine together to "[pull] out some 'coherent thread of text' from the 'statistics of conventional wisdom' that it's accumulated"²⁵. In short, LLMs are sophisticated programs which process, store, and distribute knowledge stochastically.

What makes LLMs a significant advancement over other developments in AI research is that they can adequately perform tasks that were once thought to be impossible to automate. Industries and professions which relied entirely on intellectual labour such as insurance, marketing, and journalism, can now be, partially or entirely, cheaply immitated by computer programs. Customer service interactions are dominated by robotic almost-humans as many call center jobs are easily replaced by transformer-backed speech-focused LLMs that are capable of rudimentary call-and-response communication. These programs have reached the point where they would easily pass the simplistic test thought up by the founder of computer science, Alan Turing.

Turing proposed the "imitation game" in his 1950 paper Computing Machinery and Intelligence. In it he described a simple game involving three people: a man (player A) a woman (player B) and an interrogator of any gender. The interrogator cannot see the man or woman and can only communicate by passing notes. He must correctly guess both player's gender. Player A's goal is to trick the interrogator, player B's goal is to help him. He then proffered a variant of the game where the goal is not to determine gender, but instead to determine whether player A or B is a computer. Where many misinterpret the "imitation game" is that Turing meant for it to replace the question "Can machines think?", not have the results of the game answer the question at all. This was because Turing felt that "intelligence" was not something that could be calculated or well-defined. In fact, he wrote "the idea of 'intelligence' is itself emotional rather than mathematical."²⁶ It takes little imagination to construct a scenario where a reasonably powerful LLM would breezily summersault through that trial. As a result, it is no wonder that we can now find article after article describing the harrowing tales of young adults "falling in love" with LLM powered AI avatars.

As evidenced by the fact Turing cooked up his thought expirment in the 1950s, many people have been speculating about the "humanness" or "human capabilities" of machines. Weizenbaum extensively recounts the experience of multiple people who interacted with his ELIZA program. ELIZA was intended to show that a computer program could be written to simulate human conversation. The program would take input in the form of a sentence and would then respond in kind. The most popular subroutine was titled "DOCTOR", and it simulated the sort of responses a sorry excuse for a therapist would lazily let spill out of their mouth. Weizenbaum gives a typical, real-world exchange in Computer Power²⁷:

Men are all alike.

IN WHAT WAY

They're always buggin us about something or other.

CAN YOU THINK OF A SPECIFIC EXAMPLE

Well, my boyfriend made me come here.

YOUR BOYFRIEND MADE YOU COME HERE

He says I'm depressed much of the time.

I AM SORRY TO HEAR YOU ARE DEPRESSED

It's true. I am unhappy.

DO YOU THINK COMING HERE WILL HELP YOU NOT BE UNHAPPY

I need some help, that much seems certain.

…

and so on. The responses might seem silly to the contemporary reader, but in 1966 this was considered impressive enough that practicing psychiatrists at the time genuinely believed that DOCTOR could become a replacement for psychotherapy. This horrified Weizenbaum. He saw ELIZA as a program which demonstrated the ability for computers to process natural language, he did not see it as a stand-in for real, human interaction. He quotes a psychologist, Dr. Kenneth Colby, as saying, "A human therapist can be viewed as an information processor and decision maker with a set of decision rules…" To which Weizenbaum responds, "What can the psychiatrist's image of his patient be when he sees himself, as therapist, not as an engaged human being acting as a healer, but as an information processor following rules, etc.?"²⁸ The moment the therapist is abstracted as an "information processor" she is topologically flattened into a machine—she can do no more than a computer because she is conceptually framed as a computer.

Weizenbaum does not begrudge the layman, or even the specialist, for anthropomorphizing his program. He muses that, "the fact individuals bind themselves with strong emotional ties to machines ought not in itself to be surprising. The instruments man uses become, after all, extensions of his body…his instruments become literally part of him and modify him, and thus alter the basis of his affective relationship to himself."²⁹ He later adds, "they can explain the computer's intellectual feats only by bringing to bear the single analogy available to them, that is, their model of their own capacity to think."³⁰ The machine appears to exhibit the end point of human thought (the words themselves) and, as such, must be thinking. This in and of itself is not so grotesque a notion, but the distortions come the moment the computer intelligence evangilists begin to insist that computers have all of the available faculties to replace traditional human thinking. Because the machine appears to be thinking, it must be, and since it can think, it is also worthy of passing judgment. Implicit in these syllogisms is that if the computer is thinking it is thinking like a human being.

Ludwig Wittgenstein famously stated that "if a lion could speak, we could not understand him."³¹ This is because language is a decidedly human endeavor. It is wrapped up in thousands of years of socio-material relations. Weizenbaum echoes Wittgenstein when he writes, "The human use of language manifests human memory. And that is a quite different thing than the store of the computer, which has be anthropomorphized into 'memory.' The former gives rise to hopes and fears, for example. It is hard to see what it could mean to say that a computer hopes."³² A machine is not human, so how could it possibly make sense of the subtleties which escape the grasp of gradient descent algorithms and back propogation? A machine does not feel hope. It can stochastically approximate the language of hope as laid out in billions of words funneled from Wikipedia and Project Gutenberg, but it cannot feel it. As Weizenbaum says, "Man faces problems no machine could possibly be made to face. Man is not a machine…although man most certainly processes information, he does not necessarily process it in the way computers do."³³

If a machine cannot be proven to think like a human being does, why then should it be able to govern the actions of human beings? A chilling study of AI used in place of human judgment comes in the case of the ongoing genocide perpetrated by Israel on the Palestinian people. In 2024, +972 Magazine and Local Call revealed that the "elite Israeli intelligence unit 8200…developed an artificial intelligence-based program known as 'Lavender'…which played a central role in the unprecedented bombing of Palestinians."³⁴ The AI program marked around 37,000 Palestinians as suspected "Hamas militants," most of whom were children, for potential targeting. According to the reporting by +972, "its influence on the military’s operations was such that they essentially treated the outputs of the AI machine 'as if it were a human decision.'" It was well known that the AI had an error rate, as defined by the IDF, of 90%, meaning that military operatives knew it would be "wrong" 10% of the time and did not care. One would hope that the reader ingesting this information is filled with abject horror and disgust. If one could be granted the ability to speculate it is clear the tool was used for two reasons: (1) it legitimized targets because the final results could be justified behind highly sophisticated "artificial intelligence" technology; (2) it provided the ability to offload moral responsibility as it was no longer a human choice.

Weizenbaum rigorously interrogates the first point in Computer Power: "A computer's successful performance is often taken as evidence that it or its programmer understand a theory of its performance. Such an inference is unnecessary and, more often than not is quite mistaken."³⁵ Although LLMs give off the appearance of hyper-competance, many practitioners do not understand how they work. In fact, no one really understands why LLMs work as well as they do. This is the great open secret of AI research discourse—a researcher can explain what operations such-and-such optimization algorithm performs, or which hyperparameters can be twisted to generate this-and-that output, but none of them can give a satisfying causal description as to how it is the case that a program can mimic human speech so well. Most prefer to take the path of least resistance, leading to self-flattery and congratulations for a job well done. Therein lies the psychological X-factor which tickles the minds of programmers, the great swamis of our day, and allows them to float on by, unbothered by the many cruelties directly resulting from their dogged determination to turn the world into the plaything of AI programs.

LLMs and LLM "agents", a type of LLM program which can interact with and launch other programs, have hit a level of sophistication that they are now putting many computer programmers out of a job. Why then are so many programmers and computer scientists obsessed with spawning more and more of these creations? One possible answer is proferred by Weizenbaum in his psychological profile of the "compulsive programmer"³⁶:

"The computer programmer, however, is a creator of universes for which he alone is the lawgiver. So, of course, is the designer of any game. But universes of virtually unlimited complexity can be created in the form of computer programs. Moreover, and this is the crucial point, systems so formulated and elaborated act out their programmed scripts. They compliantly obey their laws and vividly exhibit their obedient behavior. No playwright, no stage director, no emperor, however powerful, has ever exercised such absolute authority to arrange a stage or a field of battle and to command such unswervingly dutiful actors or troops."

The programmer is at a unique time in history. She exists in a moment where entire worlds are at her disposal, worlds that she alone creates and influences. They flash in and out of existence in a moment. They are constructed entirely of digitized information. "The engineer can resign himself to the truth that there are some things he doesn't know. But the programmer moves in a world entirely of his own making. The computer challenges his power, not his knowledge."³⁷ The act can create solipsistic creatures: "A solitary programmer writing a program to solve some problem incidental to his private research need hardly concern himself over the way his program might structure the worldview of anyone besides himself."³⁸ Someone who could arguably be called the first computer programmer, Charles Babbage, had the following to say after completing designs for his Analytical Engine: "It seems that all of the conditions that allow a finite machine to carry out an unlimited number of calculations have been fulfilled by the Analytical Engine…I have converted infinite space which was required by the conditions of the problem into infinite time."

Sad, antisocial thoughts pepper the worldview of the computer programmer. A trip to one of the most popular forums for programming related technical discussions, hackernews.com, will show this to be the case. As of September 11, 2026, the second most upvoted post of the past month is a discussion of the release of OpenAI's new GPT-6 Astra model. If we limit ourselves to the top comment chain we can see posts which demean human thought and reduce the world down into quantifiable categories. For example, one user states: "That's what the AI's really are terrible at – creativity. But I'd argue the vast majority of humans aren't very creative, with truly out-of-the-box ideas."³⁹ The very technically proficient AI program is not creative because human beings are not creative—the program's potential is limited by the constraints imparted by the dregs of society. Another user defends mankind's ability to think independently by saying that unlike AI programs "humans are not prompted"⁴⁰ and is quickly met with a rejoinder that "humans are constantly prompted by The Joneses and perceived authority figures (boss, religion, politics, peers, co-workers, influencers, et al.)."⁴¹ Of course! Much as a program has inputs and outputs, so too does a person. They are "prompted" by their bosses to do work in the same way that we prompt an AI to solve one problem or another—failing to recognize a person can have other thoughts in their head at the same time. Finally, one user begs of another to "define novel intelligence in a way that would not exclude 95% of humans, yourself included."⁴² Again, we are met with a typically misanthropic computer programmer. "95%" of humanity is incapable of "novel" thought, presupposing that the "novelty" could be rigorously analyzed beforehand. And so on and so on.

The act of reducing real-world problems down into what are essentially logic puzzles attracts a certain sort of person. It is a person who has an incredible aptitude for abstract thinking and problem solving. It is a person who understands how to break down phenomena into its component parts and rearrange those parts until they can be made computable. After meticulously explaining the inner workings of ChatGPT, Stephen Wolfram writes, "[ChatGPT] suggests something that’s at least scientifically very important: that human language (and the patterns of thinking behind it) are somehow simpler and more 'law like' in their structure than we thought"⁴³—our programs have stumbled onto a world beyond our comprehension, that not even philosophers could theorize. Necessarily, it is exactly this sort of attitude which forces one down a road of complete and total social isolation. The totality of being, the understanding of the human being as a member of a deeply interconnected social web, is eschewed by a practical philosophy that must strip everything down into terms which can be computed by a typed lambda calculus. "Feelings" and the intangibles which cannot be digitized are therefore to be tossed aside and relegated to the dustbin of history. Make way, make way.

In the past few years this bowling over has come in the form of many tired and predictable discussions plastered in public forums about the coming storm of so-called "AGI", or Artificial General Intelligence. This amorphous, science-fiction concept stems from the belief that AI programs will get so incredibly advanced that they will outstretch human intelligence. The thinking goes as follows: computers could not mimic human speech, then they could, but not very well, then they could reasonably well, and now they can exceptionally well. This must mean they will surpass human speech and go beyond cognitive barriers. This line of thinking is reminiscent of the suggestion of Dr. Richard Price, which Marx references in Capital Vol. 3, that if one had invested the equivalent of a shilling at the time of Christ's birth then one, in 1772, would have more coins "than the whole solar system could hold"⁴⁴. The world of science and human development is not one large geometrical progression.

Weizenbaum rejects the question outright⁴⁵:

"Does our inability to compute an upper bound on machine intelligence provide grounds either for the 'optimistic' conclusion that 'machines may surpass us in general intelligence' or for that very same 'pessimistic' conclusion? Neither. We learn instead that any argument that calls for such conclusion, or for its denial, is itself ill-framed and therefore sterile."

These programs do not think like human beings and as such they do not wield comparable intelligence. And yet, there is a non-stop flood of digital ink constantly flowing to quote the nearest AI blowhard about its coming arrival⁴⁶. Why is this the case? The answer is simple: it is in the best interest of the AI crowd to massage the public consciousness into believing that their technology is on the precipice of greatness.

As we saw above, there are trillions of dollars invested in AI technologies. Each firm is racing as fast as possible to create the biggest, fastest, and most comprehensive AI tools possible. Or, to reframe the situation, each firm must appear as if they are in the process of creating the biggest, fastest, and most comprehensive AI tools possible. Appearances are crucial to juicing further leverage and generating investment from private banks and investment firms. The AI business is expensive, but, by that very same virtue, it is also drenched in liquidity. Whoever looks to be the closest to solving the biggest problem of our day are the surest bet for the "inevitable" AGI payoff.

This "inevitability" is wielded like a cudgel which bashes the brains out of local community organizers who push back on the installation of giant power hungry data centers which blight landscapes, drain water reserves, and pollute the air with dust and unbearable noise⁴⁷. Weizenbaum asked the same sorts of questions many people affected by these technologies ask today: "Why should we want to undertake this task at all?" And, strikingly, Weizenbaum was met with the same replies that one hears today, "The most cheerful answer I have been able to get is that it will help physicians record their medical notes and then translate these notes into action more efficiently. Of course, anything that has any ostensible connection to medicine is automatically considered good."⁴⁸ The AI acolytes immediately attach themselves to the most noble and heroic of their programs' potential applications. They are smart to ignore the fact that their tools often result in rampant unemployment and are easily integrated into military technologies.

Discussions of AGI will sometimes have the AI business perform a sort of mea culpa. Sam Altman recently claimed that OpenAI will finally reach AGI by the end of the year⁴⁹. This is after he has gone on record many times about how "scared" he is of AI and its potential abuses⁵⁰. Unsurprisingly, Weizenbaum predicts exactly this sort of behaviour 50 years prior: "The real message of such typical essays [about dangers regarding technology] is therefore that the expert will take care of everything, even of the problems he himself creates. He needs more money. That always. But he reassures a public that does not want to know anyway."⁵¹ In other words men like Sam Altman are saying: "I am making the world worse, but don't worry. If you give me more money, I will continue to make it worse."

Weizenbaum passionately takes the reader by both shoulders and makes it clear that "these things are not products of anonymous forces. They are the products of groups of men who have agreed among themselves that this pollution of the consciousness of the people serves their purposes."⁵² It is of paramount importance to keep in mind that people have chosen the world to be like this. The wind does not blow and construct a data center. The clouds do not burst and install Google Gemini on your child's Chromebook. In fact, it was "men just like the ones who design television commercials [who] sat around a table and chose."⁵³ Weizenbaum calls the rhetoric of inevitability "a powerful tranquilizer of the conscience"⁵⁴. He continues:

Its service is to remove responsibility from the shoulders of everyone who truly believes in it.

But, in fact, there are actors!

For example, a planning paper circulated to the faculty and staff by the directors of a major computer laboratory of a major university speaks as follows.

'Most of our research as been supported, and probably will continue to be supported, by the Government of the United States, the Department of Defense in particular. The Department of Defense, as well as other agencies of our government, is engaged in the development and operation of complex systems that have a very great destructive potential and that, increasingly, are commanded and controlled through digital computers. These systems are responsible, in large part, for the maintenance of what peace and stability there is in the world, and at the same time they are capable of unleashing destruction of a scale that is almost impossible for man to comprehend'

Responsibility is offloaded to "systems". The laboratory cannot do anything but get on with the times. The famous refrain of "if I won't do it, someone else will". Anyone who says this when their work is actively engaged in the destruction of innocents is a coward and profoundly morally compromised. AI dominance is not "inevitable". It is only inevitable if the world rolls over and plays dead.

To restrict the sandbox in which AI programs can be let loose in is not to revert to "mystical" ways of thinking. I am staunchly of the belief that thought is material. We live in a physical world and think physical thoughts. It is wrong, however, to believe that an impossibly large, arbitrary collection of neural networks can be understood to think the same way we do. As Nietzsche understood, our world is human, all too human. There is no obvious advantage to offloading vital cognitive functions to programs whose innerworkings most people barely understand.

Weizenbaum's greatest fear was realized less than two decades following his death. America is peppered with advertisements for AI-powered therapy applications designed to streamline the mental healthcare process to turning on and off one's phone. The newly christened Department of War makes backroom deals with major AI "hyperscalers" to inject LLMs into the global imperial war effort. Education is teetering on the edge of being dismantled and replaced by chatbots programmed to spit back encouragement to an antisocial student body bombarded at all times with short form video content and manosphere influencers. The world has rushed head first into a digital age whose fists are dripping with the blood of pulverized enemies who dare to stand in the way of progress.

There is only one solution to this total domination: revolutionary political action. Capital has no incentive to pump the breaks on AI expenditure and investment. So long as Wall Street, the IMF, and the World Bank believe AI to be the future, then AI will be the future. The only winner can be whoever decides not to play the game. What remains to be seen is if there is a likely challenger in the dugout, waiting to dust of their cleats and take the mound.