AI and the destruction of thought

How GPT AI impoverishes discourse and culture

I’m going to go out on a limb here and for the sake of argument assume that it’s possible to speak of “progress” (in everything from science, philsophy, the arts and politics) in a meaningful sense.

We can use a quantitative notion (the dominant modern perspective) where knowledge accumulates and sometimes gets refined by new theories or observations that build upon said accumulated base.

We could also assume a qualitative notion of progress, sometimes expressed in relation to the arts, maybe more often in politics, where positive developments involve the change of not only the theoretical apparatus - but even of the very preconditions for interpretation and of the values that guide our judgments.

Progress here means transcending previous theoretical bases to discover entirely new forms of knowledge.

A third perspective which isn’t so much a position on progress per se, but which I find much less authoritarian than either of the above, is to jettison the entire concept of preferable, universally prescribed goals for society and just let people decide for themselves.

Let the Spirit blow wherever it will.

+++

Anyway, the point is that progress in either of the first two senses (and definitely the anti-progress alternative) will be effectively undermined by the ascendancy of GPT AI technology and similar tools.

Why?

So the development of cumulative knowledge within a self-correcting framework (as per the contemporary myth of modern science) presupposes a constant juxtaposition between dominant perspectives and alternative points of view. This is how knowledge is improved, how theories are challenged and added to, and how new paradigms capable of discovering and describing novel facts are developed.

You need the “wrong-thinking” outsiders even for merely quantitative progress, since new knowledge within established paradigms presupposes that somebody asks questions that aren’t already implicit within the dominant perspectives, narratives and modes of interpretation.

You need these challenges if for no other reason than that they provide incentives to find new ways to defend the established positions.

Even St. Augustine, a representative of the oh-so-dogmatic Catholic Church, once maintained that the reflected defense of heresies ultimately contributes to the flourishing of human knowledge, something which not least is manifest in St. Thomas Aquinas’ conscientious fortifying of the positions of his adversaries before he advances his own arguments.

And you most definitely need iconoclasts for the transition into new paradigms (new theoretical apparatuses, new ways to frame facts), since this presupposes that someone advances theories that are actually negated by the dominant points of view.

Theories that, legitimately or not, are considered false, given the assumptions inherent to the established paradigm.

So progress in the merely quantitative sense actually requires contrarian points of view. And the more important the matter, the greater the need for the corrective influence of a pluralistic and multifarious set of often opposing viewpoints.

Today, this is generally known as “disinformation”.

The qualitative notion of progress is even more dependent upon contrasting views to be realized in practice. While quantitative progress can chug along without dissent for a while, even if it’s ultimately stymied, the qualitative version doesn’t even get off the ground.

The radical reinterpretations necessary for the disruptive and radical paradigm shifts ushering in entirely new ways to discover the world are namely one and the same thing as dissent from the established consensus. And this holds whether or not this consensus is an enforced monolithic one, or merely a set of mutually incompatible, competing perspectives.

If you’re going to add something “new” - if you’re going to achieve qualitative progress in terms of a radical paradigm shift unlocking new facts and bases for theory and practice - you need to advance theories and factual claims that are falsified by the dominant perspectives and worldviews (that is, not positions effectively falsified in the epistemic sense, but such which would be considered false if we take these established positions at face value).

You need to assault the “fact checkers”.

All right. So why are technologies like GPT AI a problem in this?

Because it’s essentially a framework for giving us authoritative canned answers. For the reproduction of entrenched paradigms as a substitute for actual exploration, research and discovery.

Google and Wikipedia are bad enough. These platforms are the most effective tool for an enforced consensus throughout every sphere of human knowledge since the development of authoritarian, proprietary mass media.

The GPT AI supercharges this, however.

The first reason is that it intimately connects with human psychology because it actually seems like an authoritative agential entity. Another conscious subject. This also receives a force multiplier from the scientistic myths of progress and technology.

So you ask the AI questions, and you receive the CREAM OF THE CROP of all the knowledge imbued within spectacular modern science and technology.

Unlike the humble search engine, powerful as it is, the GPT tech doesn’t give you a list of results of alternative and possibly competing sources of information, however curated such a list may be. It provides a condensed answer modelled after natural human language use, generating a mimic of an omniscient oracle. The kids in the quote above? 50% of their feedback responses detailed their shock that the GPT tech could be misleading.

That’s quite a level of implicit trust.

Also, it’s proprietary, of course.

“If anything in world history could aggregate economic power into a small number of hands, it is unambiguously AI,” he said. “If you got into the game now, you’d need $100bln, maybe $200bln, and your odds of winning would be no better than 20%.” Winners will keep sucking earnings from the economy. “Adapt your business to this trend. In real estate, build beds for machines. That’s what a data center is. Get short humans, long machines. People say buy Austin real estate, buy Miami, but that misses the opportunity. Find ways to buy Machinistan.”

ZeroHedge

So why do you think big tech throws two hundred billion dollars into AI? To provide for the most diverse number of perspectives on any given issue towards maximizing mankind’s intellectual richness and liberty?

Or is it for profit, maybe?

Do you think “history’s most intense aggregation of economic power into a small number of hands” sounds compatible with a proliferation of contrarian perspectives?

Yeah.

GPT AI and related technologies will by their very structure tend to reproduce the dominant positions in every single field, regarding every single issue. Since it must operate by parsing patterns in available data, it will inevitably mirror the most influential perspectives, theories and narratives and MASSIVELY reproduce them all over the digital mass media.

And this of course means that the set of ideologies, traditions of knowledge or body of theories that become dominant in this brave new digital framework can potentially attain something like full spectrum dominance over knowledge production.

Add to all of this the automated suppression of politically incorrect ideas and the AI-governed restriction of undesirable content in conjunction with massive surveillance and social credit analogues and we’re so far up the dystopian shit creek that even the Victory Gin has run out.

Reclaimthenet

This is the end of liberal politics. The death-knell of science. This is the death of reason.

This is the entrenchment of an immutable soft totalitarianism, totally immune from anything akin to democratic supervision or rectification for as long as the technological infrastructure still remains. This line of development, taken not much further, will render not only discourse, but the very fundamentals of thought itself utterly impoverished.

It will eliminate whatever vestiges of diversity that remain within human traditions of knowledge, and will leave us utterly defenseless in the face of even the most blatant of tyrannies, since we will no longer have the necessary conceptual tools to identify it and mount a resistance.

The relevance of abstract questions, the content of the answers given, the quality of life adumbrated in these answers - all these things can be decided only if everyone is permitted to participate in the debate and encouraged to give her or his views on the matter.

The best and simplest outline of the ideas just explained is found in Protagoras's great speech (Plato, Protagoras, 320c-328d): the citizens of Athens do not need any instruction in their language, in the practice of justice, in the treatment of experts (warlords, architects, navigators); having grown up in an open society where learning is direct and not mediated and disturbed by educators, they learned all these things from scratch.

As for the further objection that states and citizens' intitiatives do not arise out of the blue but must be set in motion by purposeful action — that is easy to answer: let the objector start a citizens' initiative, and he will soon find what he needs, what furthers his ambitions, what obstructs them, to what extent his ideas are a help to others, to what extent they hinder them, and so on.

Feyerabend, Farewell to Reason