Imitative AI and the Problem of What Problem You are Solving

I hate boilerplate code — the code that handles all the little plumbing details, the setup, the config, the weird syntactic quirks of each language etc., the rote and routine code that every stable, production ready program/system needs or needs to adjust to in order to run. I hate it the way healthy people hate the plague or rich people hate paying people for their work. The first thing I always did at a new job or with a code base unfamiliar to me was to write code to write the boilerplate for me. Boilerplate code is a problem, one that imitative AI plausibly solves. But it might be the only problem that imitative AI really does solve on an enterprise, and thus at a commercially viable, level.

There are only two things that matter in commerce: what problem am I solving and how much is solving hat problem worth to people with money. I know that in modern capitalism it may not seem that way given how much of economic activity centers around consumption, but a lot of people genuinely believe they have a version of the “I’m not cool enough” or “Not enough people know I am cool and rich” problem. Welcome to capitalism. Regardless of why people think this way, they do, and solving those problems is a massive part of our economy. Unfortunately for imitative AI, it cannot solve that problem. Even more unfortunately for imitative AI, it really cannot solve many problems.

Most imitative AI, in a business context, is used to summarize and write reports and emails, replace front line customer service, and do coding realted tasks. Unfortunately for imitative AI, that is about the list of where it is effective. Perhaps that changes, but it also doesn’t change the fact that imitative AI is highly, deeply subsidized. The real costs of its use make it, even coding, an iffy economic proposition. And the attempts to find a problem to solve often just prove that imitative AI doesn’t really solve problems.

OpenAI shut down its movie/animation generation program Sora because it wasn’t making enough money to justify its continued existence. Sora was meant to solve the “problem” of paying people for artistic work, but it failed. Another example is the plan to allow imitative AI chatbots to prescribe psychiatric medicines. Now, this actually sounds worse than it really is. The program is limited to certain low risk medicines and require an ongoing prescription, which it cannot generate itself (though given how these thing make up bullshit, I do wonder how they intend to keep the system from deciding for itself that a prescription exists), patients who are not stable cannot participate, and the program requires regular checkins with a clinician. This, by the way, is exactly the process that happens today — you see a clinician, they prescribe with a certain number of refills, you call the pharmacist when the refill is needed. If you don’t regularly see a clinician, you will eventually not be able to refill. What problem, then, is this supposed to solve?

It doesn’t solve the lack of clinicians, since people are supposed to still see their clinicians. It doesn’t solve a speed bottle neck, as most people can get refills automatically on most of these medicines form pharmacists already. it doesn’t make the meds any cheaper — costs is not a function of clinician spending but rather market and monopoly forces within the pharmaceutical industry. It exists entirely because some elected doofus didn’t want to be “left behind” and made someone in state government do something publicly facing with AI, because its “the future”. Like most imitative AI, this system doesn’t solve a problem.

Which is, well, the problem with imitative AI in an economic sense and an example of how our economics are broken. It is clear now that there are very few problems imitative AI solves, either because they cannot be trusted or because they cost too much. Even where they do provide value, like coding, the value largely exists because they hide the true cost of the product. In any sane system, these investments would have already been written off. But we don’t live in a sane system.

Our economics are largely controlled by monopoly forms and the investors who profited from the creation of those monopolies. And those people want tow things: to repeat the easy wins of the early internet and social media ages and to have absolute control over society. Imitative AI, if it really could replace human work on a mass scale, would provide them both. That it cannot is not recognized as a truth but merely as a temporary inconvenience. Power and wealth beyond their dreams of avarice are waiting for them at the end of the imitative AI rainbow. And because our economy is dominated by monopolies, they can push their dreams far past the point of sanity.

I guess, then, the problem that imitative AI is trying to solve, at least as far as the owners are concerned, is the problem of having to live within a democratic society. This is why the continually push it into government and education, that is why they are so desperate to see it work, where work means taking over the economy. The question now is: can they infect enough of our society before the economics and product liabilities come crashing down around them?