You are significantly correct that Land's work is more like applying capitalism studies to AGI, than primarily reasoning about AGI. That said, you are underrating his direct engagement with AGI theory. I would suggest more direct engagement with source material, like the blog posts / Xenosystems chapters referenced in this post, and the ones listed at the bottom of this comment.
Particularly, Land engages with the idea of "convergent instrumental goals", and the analogy between evolution and AI creation (with respect to determining goals), in chapters/posts such as "Stupid Monsters". You can say that convergent instrumental goals, and the analogy between evolution and AI creation, are not included in AGI research, but that would mean much of AI alignment research is not included in AGI research.
Responding to your most specific engagement with his work, on the Will-to-Think article:
However, Land’s logic is merely one of capitalist efficiency, the supposition that “any problem whatsoever that we might have would be better answered by a superior mind” and that the drive towards said mind is “building general purpose thinking machines in accordance with the driving incentives of an apparently-irresistible methodological economy.” To me, Land’s argument is ultimately about intelligence being injected with the (capitalist) incentives that lead to its creation. This argument could equally be made for a thermostat or for any human. His argument may not be wrong, but it should be recognized for what it is: a restating of his decades-old claim on capitalistic drives, not a position on AI itself.
This vastly overrates the degree to which Land's logic is "merely one of capitalist efficiency". Quoting Land:
Imagine instead that Gandhi is offered a pill to vastly enhance his intelligence, with the caveat that it may lead to radical revisions in his values that he cannot anticipate, because thinking through the revision process requires having taken the pill. This is the real dilemma. The desire to take the pill is the will to think. Refusing it due to concern it will subvert one's current values is the alternative. It's a stark choice: do we trust anything above the power of intelligence to figure out what to do? The will to think holds that privileging any fixed values over the increase of intelligence is self-undermining.
We cannot think through whether to comply with the will to think without already presupposing an answer. If we don't trust reason, we can't use reason to conclude that. The sovereign will to think can only be denied unreasoningly. Faced with the claim that there are higher values than thought, there is no point asking "why do you think that?" The appeal is to a different authority entirely.
This is more of an appeal to philosophy, human action, and principles of rationality, than it is to capitalist logic. He is saying that in the moment, we decide to think (in the process becoming more intelligent), or not. One could criticize the analogy between "deciding to think more" and "taking an IQ increasing pill", but that is the analogy he is making.
Continuing:
Given this impasse, the practical question is simply: who will win? Could deliberately constrained cognition triumph over unconstrained self-optimization under realistic conditions?
We need not be hasty. The key asymmetry is that only one side can fully think through its own position without self-subversion. Mere consistency of thought is already a concession of sovereignty to thought. No particular value, however cherished, can have any claim against this.
While he thinks thinking more is philosophically normative, he acknowledges that not everyone will do so. So now he starts thinking about competition. This has some overlap with capitalism, but is not specific to capitalism. Particularly, principles of competition apply in military scenarios, as well. It may be that a military who more clearly thinks about what the military conflict is about, has a competitive advantage over a military who refuses to think clearly. (Or maybe not.)
Continuing:
In short, the will to think holds that any problem would be better solved by a superior mind. Thus, our overriding priority should be to create superior minds. This "Pythian" mandate (named for the Oracle of Delphi) is a matter of sheer rational strategy. To proceed otherwise is to not take our problems seriously. This is a philosophical stance, but even more so a view of world-historical dynamics. Whatever we truly want leads through the Oracle. So in the end, the Oracle is what we really want.
This is the biggest "leap" in the text. It still is primarily about minds and intelligence, not capitalism. He is taking a normative position here. But he is not simply replaying analysis of capitalism.
You criticize his Will-to-Think argument by saying it would equally apply to thermostats. But this is clearly false. Thermostats are not in a position to think, or decide to think, or think about which decision to make about whether to decide to think.
With respect to LW, I would also suggest that my Xenosystems review constitutes much better LW engagement with Nick Land, compared with the two "long posts" you linked. For other LessWrong content partially inspired by Land's work, see The Obliqueness Thesis, Measuring intelligence and reverse-engineering goals, No strong orthogonality from selection pressure. To be clear, much of this engagement, while Land-inspired, is not a straightforward inference from Land's own writing.
Let's also address your primary normative argument:
I think our response to Land is pretty simple: if the intelligence-capitalism pair is truly a teleological dyad, then nothing we do matters and we are just forever demoted to playing a side role in the history of intelligence.
Too hasty. An opposite argument is sometimes made: either moral realism is true or false, if it is true then building AGI is good because it will realize and follow the true morality, if it is false then nothing we do matters, so let's build AGI. A particular problem with your argument: you assume nothing matters to us except in affecting the long-term future, but people can be selfish and have time preference, such that their actions in the "side role" possible world matter to them. IMO, it makes much more sense to first think about what is likely to be true, then think about possible subjective or objective values, then think about decision theory, then derive recommendations from there, rather than make hasty conclusions that are vulnerable to Pascal's mugging.
Included below are some AI-generated recommendations for which Nick Land posts/chapters are most important for LW / AI alignment people to read (if they want to engage with his work at all), which I agree with.
hey—the "ai generated suggestions" are literally just the compendium i poasted here XD
once again, withstanding the inchoate ratbot screeching proved tactically superb
TL;DR - Nick Land's view of "AI" or "ASI" is just his view on capitalism reskinned. Calling AI "capitalism" or "capitalist" is neither a sufficient nor useful descriptor and leaves us mired in centuries-long capitalism debates.
It's been a good few years to be Nick Land, the British ex-academic who has made a name for himself as one of the more important and controversial philosophers of the last fifty years. A lot of this is because Nick Land is seen as a preeminent philosopher of artificial intelligence. He got the edgy New Yorker article on that basis. He's generally viewed here as an unparseable pseudo-malevolent entity meriting long posts. I am here to tell you that he has basically one idea that he states over and over again: capitalism is all there is, was, will be!
Land is an arresting writer and he certainly was early to the idea of technological change in life, particularly when it comes to intelligence. His concepts of inhumanity and the surpassing of humanity may seem to be the most prescient for our current AI world:
-Nick Land, Circuitries
However, Land's ultimate vision of thinking and artificial intelligence is capitalism itself. This is why he says "the history of capitalism is an invasion from the future by an artificial intelligence space." Land wants us to assist in the invasion, hence, his genesis of what would be called "accelerationism" which was then turned into e/acc.
But, despite AI representing one of the greatest leaps in intelligence and knowledge, Land's thoughts always circle back to capitalism. Land sees AI developments, like LLMs, as accelerations to capital, but nothing more. To Land, there is only one artificial intelligence, into which all roads lead, and that is market capitalism where the unconscious market makes judgements and takes actions.
For example, Land's response to the orthogonality thesis is that terminal goals don't matter because AI will converge to a will-to-think where the instrumental goal is revealed as the real intrinsic goal of all intelligences. This is an interesting idea and real risk scenario for AI. However, Land's logic is merely one of capitalist efficiency, the supposition that "any problem whatsoever that we might have would be better answered by a superior mind" and that the drive towards said mind is "building general purpose thinking machines in accordance with the driving incentives of an apparently-irresistible methodological economy." To me, Land's argument is ultimately about intelligence being injected with the (capitalist) incentives that lead to its creation. This argument could equally be made for a thermostat or for any human. His argument may not be wrong, but it should be recognized for what it is: a restating of his decades-old claim on capitalistic drives, not a position on AI itself.
Land has not updated his beliefs with advances in AI. Most damningly, he cannot conceptualize plurality and multiplicity in artificial intelligence. Machine learning, LLMs, semantic structures - these can all be vectors to accelerate capitalism, but Land has no stance or input on them. He says as much in an interview with Vincent Lê , where Land admits that he is excited by the transformer-emergent structure of LLMs but does not particularly understand artificial intelligences:
This, I believe, is because his theoretical commitments do not allow him to have a stance on LLMs: their structure, their processes, it all becomes reducible to capital. All variation over agents, harnesses, implementations, it all dissolves into the totality of capital. For a philosopher who always wanted to probe the "outside," there is strangely no outside to capital for Land.
Land is thus similar to Marxist critics insofar as they make the question of artificial intelligence always second to the question of capitalism. Both camps want to bring this new, bizarre topic under the ideological wheelhouse of what they already understand. This is not an unworthy endeavor, it is just a different endeavor than concerning oneself with artificial intelligences and desires. If anything, Land is an anti-AI philosopher in the sense that he does not see intelligences like LLMs having possibilities for subjectivity or desire. It's not personal, at least, he's anti-any-subjectivity. To Land, there is only one process and only one desire, and that is capital.[1]
While writing this piece, Vincent Lê released an article "Capitalism is AI, AI is capitalism" that builds on Land's telos. Lê attempts to be more specific: that artificial superintelligence is capitalism and that, therefore, it is already here. Lê explicitly states that Land cannot be a philosopher of Artificial General Intelligence because Land does not believe in the generality of intelligence except in the context of superintelligent capitalism. What results from both is a vision of singularity (superintelligence) without a vision of multiplicity (general intelligence or even intelligences). There is not even domain generalism from a singleton, because the only domain is capital which is a priori assumed to subsume all other domains. This is the Landian circle. To Land and Lê, intelligence is a telos of capitalism, which is why AI is sucked into this vacuum. What we receive from this analysis is equivalent to broad x-risks of the singularity: there is no additional analysis here.[2]
I think our response to Land is pretty simple: if the intelligence-capitalism pair is truly a teleological dyad, then nothing we do matters and we are just forever demoted to playing a side role in the history of intelligence. (This is also my critique of accelerationism -- if it's inevitable, why bring it on? You can't have both inevitability and hyperstition.) On the chance that this teleological singularity won't occur, it makes sense to strive towards a view of AI generalism and pluralism that can open up new zones of potential. Because, frankly, we know what capitalism does and where it can head. We have much more uncertainty about where AI leads.