Purposeful joy

09 Jun 26

This blog used to be a boringly nerdy technical blog, containing entries on esoteric aspects of one or two computer languages. I purported to be enough of an expert in a few niche aspects that I felt able to publicly opine. But the value of expertise is one of the things that has been intentionally destroyed. Everybody now rushes to pay untold and unpredictable amounts of money to one or two companies for the ostensible convenience of having expertise delivered via an API. Delivered cleanly, without friction, in irresistibly self-affirming ways.

And delivered in a vacuum. Expertise used to be socially constructed, negotiated, validated. Expertise was never a given. And yet now it's just there for any and every body. Different levels dependent on how much people can or are willing to pay, but nevertheless expertise on tap, entirely devoid of any background, context, or friction.

I am no longer any more of an expert than you in any of the domains which I used to joyfully explore. Where does that leave my joy? I honestly don't know, and am writing this as a way of exploring where joy might go in the face of the hundreds of billions, potentially soon trillions of dollars worth of political power capturing every influence in almost all political, economic, cultural, and social trajectories for the foreseeable future.

Let's start small. AI was almost exclusively developed by a bunch of computer programmers, and their first or primary area of application was by intention, design, or mere short-sightedness, computer programs themselves. Granted extra impetus by the belief on the part of some with elevated decision-making powers in a sci-fi fantasy of self-programming computers. But self-programming computers need purpose before ability, and it remains entirely uncertain how purpose even tangentially relates to stochastic parrots, glorified auto-complete engines, massively over-powered pattern recognition systems, or whatever other form of AI you care to name. None of these machines "have" anything, for nothing inheres. They merely "are". There can be no discussion of inherent "purpose", and therefore no sensible discussion about self-programming computers (world-destroying paper clip factories notwithstanding).

A person can only give form and not purpose to a computer program. The form encodes purpose, but the purpose itself resides somewhere in the space between the person conceiving that purpose, and way that they translate that purpose into instructions that mean something to the computer system. Just like a book can not be meaningfully said to have any intrinsic purpose other than to be read, neither can a computer program. Both books and computer programs are (mere) manifestations of authorial purpose¹.

To consider why this might matter, let's take the typical cliche of a notional "manager" reducing a collective of human coders to one or two, so they can replace the rest with a subscription to a single commercial AI API. The manager only ever conveyed purpose in broadly general terms. The coders' collective may have had some kind of intrinsic or extrinsic hierarchy through which purpose also flowed, mostly likely gaining specificity and losing generality at it propagated downwards towards the actual people who ended up typing the code symbols on their keyboards.

Where does or did the purpose then lie? How does or did purpose change as it propagated down, and at what level can that purpose be approximately replaced by a machine that has no purpose. That is evidently equivalent to asking at what level purpose then ceases to matter? At the notional lowest level in that kind of system, the purpose of an individual coder might be considered to be the fulfilment of a specific programmatic task. The context for that task lies within their own coding community, within the existing code they have to modify or extend to fulfil their task, and within the full range of human idiosyncrasies manifest in the unique results of their coding efforts. Can people at this level be meaningfully said to manifest purpose in their labour?

That question is as old as humanity, and its answers as varied, even though many societies have converged upon notions of labour being purposeful in enabling wider ranges of social engagement (even if only though remuneration). Yet most societies which have bestowed upon us recorded ways of addressing that question were also noticeably hierarchical (shout-outs to Plato and Confucius), and the question was almost always itself posed from a top-down perspective. The value of labour was always a question for the consideration of those at least one step removed from those doing the labour themselves.

Purpose, for those posing such questions, was an abstract concept of no direct practical application to themselves. They simply had inherent purpose. The stating of such questions was an admission that the value of purpose decreased for those on lower hierarchical levels. Their purpose did not and does not matter as much. This decreased value of purpose made people more substitutable in exactly the same way it makes them more able to be replaced by AI.

So if purpose matters, it must inhere at levels higher than that of those responsible for direct typing of computer code. What about those who conceive of and design entire computer programs? What is the purpose of the programs of which they conceive? And where does that purpose come from? And most importantly, what is it about that purpose that might render it a distinct category of purpose that is not replaceable, while the purpose of the coders lower down is?

No such categorical distinction can of course be justified or made. Especially in a moment in history when so much narrative about the tools we have now is based on some envisioned and entirely hypothetical future in which anything they can't do right now is not so important because they'll be able to do it sometime soon, given enough money, energy, power. If those arguments are acceptable, it must by direct analogy be every bit as acceptable to argue that any attempts to distinguish categories of replaceable versus not replaceable on the basis of purpose are moot. Any attempt to justify some kind of replaceability gradient must be judged alongside these kinds of narratives as transient at best, and likely unstable or irrelevant in the face of any of these kinds of anticipated future developments².

The relentless narratives of need for speed, for pre-empting future trajectories, for leading and not following, all serve as arguments for pushing any notional gradient further along or up or out towards higher levels of replacability. At the top of anything like that usually sit managers. For publicly listed companies, they generally have a singular and precisely-defined purpose. And that purpose is entirely extrinsic; it in no way inheres in any individual manager, and so they are also as AI-substitutable as all others. Which leads directly to Clawbot auto-CEOs, and the death of everything.

And so we come to a conundrum: Any argument of the replaceability of labour by AI at any "sufficiently" low hierarchical level can not be categorically distinguished from any other hierarchical levels. And that leads directly to the entire substitutability of everything by AI, after which the only question that remains must be: What is then the point of any of this?

Perhaps one useful approach to figuring out what this current historical and transient phase might be "about" is to try out a few ways to ask, "What happens to ... ? Try a few words, and see which might best enrich understanding of where we are, first and foremost, and then of where we might plausibly end up.

The kind of thought experiment above can easily be applied to many words and situations. What happens to art? What happens to societies? What happens to democracies? What I want to consider first is, What happens to purpose?

Purpose is as impossible to define as consciousness, and just as subjective, dynamic, and bound to our human condition. Much of human struggle can be seen as striving for broader ascriptions of purpose. Those histories are still enacted every time words like "enigmatic" are ascribed to powerful figures. That word is rarely used to describe those forced to hold lower places in social hierarchies. The word "enigmatic" is used to disguise a belief in purpose in cases where purpose can not be discerned.

Purpose is merely more likely to be ascribed to those in perceived higher hierarchical levels. Such ascriptions are nothing other than a manifestation of historically ingrained social habit. Purpose can be found in all categories of people, and is not restricted to any domain. But ascriptions of purpose in many societies over long historical durations have tended to point upward.

Given the generally upward-pointing nature of purpose, the established and intentionally inbuilt self-reinforcing mechanisms of AI machines provide an entirely sufficient mechanism to further strengthen assumptions that purpose inheres more in higher than in lower levels. Those at higher levels will insist that, by virtue of their purpose alone, they hold the uniquely irreplaceable positions, and that they are therefore permitted to decide on which of those below them is replaceable or not.

Purpose need not inhere to any greater or lesser extent in any level at all for this mechanism to take effect. It appears inbuilt and entirely unavoidable within current manifestations of LLM-driven AI models. Purpose can remain as undefined and undefinable as ever; we need not move at all closer to understanding the nature of purpose; and yet purpose will have its way regardless. It will become the purpose of a generation to make decisions as to which of those below them should be replaced by machines. Those being replaced will simply have their purpose made void.

And so to return to joy. A phenomenon that is difficult to ascribe to a particular class of hierarchical level, joy is something more likely to have been recorded for posterity as distributed throughout all societies largely regardless of otherwise mitigating factors. What happens to this socially dispersed and untethered thing that is joy?

To understand what might happen, consider that joy is also related to purpose. And much of the post-war 20th century into this current century has arguably seen an increasing overlap between labour and purpose. The effects described above are likely to further strengthen the coupling of purpose to labour. And with that must also be presumed an increased entwining of labour, purpose, and joy.

But the exacerbation of pre-inscribed and in-trained hierarchical bias in purpose will be unlikely to apply in any similar way to joy. And so joy will be left less tainted by self-reinforcing feedback cycles of future AI models, and with that, potentially more free.

Where will then be the spaces for joy within societies profoundly affected by centralised AI infrastructure? I can only speculate, and mostly only do so out of my own personal experience of having had much of my joy effectively taken from me at the hands of auto-completecoding machines. My joy has had to refocus, to move elsewhere. I suspect and hope the same will be true of others.

Countless reams of paper have been printed in consideration of the relationships between joy, purpose, and economic necessity. Centralised AI infrastructure can not avoid prompting a new evolutionary direction in these relationships. What happens will both depend upon, and reveal, a lot about the importance of joy, purpose, and economic necessity in human endeavours. It is beholden upon any of those who believe in Sam Altman's philosophically void effusions about ages of golden prosperity to immediately move on to deeper considerations of the importance of purpose and joy in human endeavour. If planning behind AI developments purports to be truly long-term, we need to start seeing considerations of joy and purpose now.

But I don't believe any of that will happen, and I don't believe any of those who are, or ultimately will be, responsible for centralised AI infrastructure to think about anything other than becoming even more insanely rich at the expense of everybody else. Thinking about joy and purpose will be a task for everybody else. And it will have to be done absent any fabled golden age of human prosperity, which sure won't came while we're sharing our only world with trillionaires.

Of these three factors, economic necessity could go away, but there are appallingly few trajectories likely to make that happen. Purpose will be subject to the winds of economic necessity in ways likely to be very difficult to predict. And joy will be taken along for the ride, inseparably tethered to both purpose and economic necessity.

One way to view this is that we are simply in another round of Neo-Liberalism versus Marxism. The Neo-Liberals inventing words and fabulous mechanisms to continue not giving a damn about joy (does joy trickle down?) while declaring the mechanism of wealth generation to be the sole mediation and definition of purpose. And the Marxists screaming dire warnings of what happens to societies that decouple both from economic productivity.

And yet the advent of centralised AI infrastructure enhances this view in one important way. It's not just Neo-Liberals versus Marxists on opposites ends of some notionally level playing field. I have argued here that centralised AI infrastructure will exert a similarly centralising force on understandings, or at least the manifest value, of purpose. Centralised AI infrastructure will further enhance increases in the value of purpose with higher hierarchical levels, while the value of joy will remain joyfully oblivious and independent.

So the playing field is not level. As this notional system evolves from its current state, it seems unavoidable that the value of purpose for all will be determined by decreasingly few. Conversely, the value of joy will remain approximately universal, and therefore overwhelmingly in the hands of the many.

So once again it is maybe no different to an eternity of the money and power of the few versus the numeric majority. But merely identifying those things which must remain overwhelmingly in control of the majority surely helps?