Who cares about the inevitable impact of GenAI on creative employment?

The GenAI progress is astonishing in how well machines are able to replace higher order human functions. But we ought to think through the consequences of this rate of progress on human creativity and livelihoods.

Hey, so I get it — every day some new AI awesomeness is landing, I mean that is SO RAd (pun), right? AI is also my bread, I-can’t-believe-it’s-butter, and an assortment of low-carb low-fat foods. I also get excited by all this innovation — “what a time we are living in” and all that. I get it, and I appreciate it because it’s my job to know what’s cooking in the AI land. In fact, as an AI practitioner, I owe all of my full-time jobs and most of my earned dollars to various AI waves. So some might say I am crazy to be sawing off the branch I are sitting on. I still think this is worth discussing, and coming from me gives it more credibility than some non-AI dude saying “AI is taking our jobs”.

So it’s good for you and it’s progress. Why you complain, bro?

Something seems different this time around. Everyone and their uncle seems so excited about the rapid advancement in a machine’s ability to ‘create’, but there appears to be little balancing act from the other end. What happens to all the jobs that will be made redundant faster than you can spell ‘redundant’? Progress at all costs, including loss of livelihood? I don’t understand macroeconomics or politics enough to know who should be deciding if any intervention is needed— probably the governments, because trillion dollar companies building foundation models have a conflict of interest. Or perhaps some subset of the responsible AI developers in the industry and academia? Perhaps all these conversations are happening at some level, but they are not reaching us commoners.

Anyway, here are some specific line items worth considering:

1. I have a growing number of unemployed friends and family whose jobs are getting actively made redundant by some kind of ML system, or the prospect of it. And their skills aren’t easily transferring to something else.
2. For now, the job losses are focused on the lower levels, and they aren’t coming back. For now, the strategic leaders, creative heads, visionaries are not (yet) impacted by it. The volume of such roles is orders of magnitude less.
3. That’s coz foundation models learn best from extreme volumes of repetitive data drawn from naturally occurring (sequential) distributions like language and vision. The output of folks at lower levels of the food chain tend to have these properties. Now put 2 and 2 together.
4. Going by that, a founder’s job might feel safest because it’s open-ended, strategic, and with little repetition. But there’s only so many founders that a consumption-oriented economy can support.
5. Eventually, perhaps everyone will find new kinds of employment. But what happens in the interim? What the leaders are saying is mostly BS. There’s not enough honest conversation happening around what happens to these displaced jobs in the next few years.
6. Yes, we’ll have many more creative tools in their hands, but the reality is that consumption of monetizable creative production will be bounded by population (sublinear), time (constant), and bar on quality (??).
7. Competition will intensify with higher production volumes, but just like today most of that content won’t get consumed. Pareto Principle will apply more extremely — 80–20 will go to 99–1 because collective human time is bounded i.e. 99% of the content won’t get consumed.
8. Statistically speaking, you probably won’t get wealthy from all this innovation. We are heading toward a winner-takes-all market. A tiny number of trillion dollar companies (and within them only the top people) will get even wealthier. The disparity between rich and poor will grow, not shrink.
9. I think it’s inevitable that AI creations will pass Total Turing test, Ebert test, etc. So don’t get misled by claims of ‘human touch’ being eternal. The masses, who drive the economy, won’t know the difference and won’t care.

As you open that champagne celebrating humanity’s ability to automate it’s own increasingly complex abilities, don’t forget the little people, the powerless, clueless, resourceless ones struggling to find jobs and can’t keep up with all this innovation.

I hope I am wrong, but probably not. I don’t have a solution that doesn’t involve slowing down innovation, but I am hoping others do?

I wrote this on LinkedIn and got some interesting questions:

[some variant of] Machines can never place the ‘human touch’, so no need to be concerned.

Umm, very soon, I mean in months, we would have crossed the uncanny value of human likeness and reached a place where human and machine creations will become indistinguishable in a blind test. It’s inevitable; in fact it’s already happen for 30–40 of all machine creations. So this argument will just not hold for long. For similar reasons, Marc Andreessen and Ben Horwitz said in a recent podcast episode that any attempt for machines to decide if a student has used an LLM to write an essay would fail. I suspect that in less than a year, such a system will do no better than random chance at telling that difference human creation and the top LLMs.

Isn’t it just like the industrial revolution? It’ll all work out in the end.

I am sure there are many many differences, and I am no expert on this topic, but the one that stands out for me: the industrial revolution brought tools and techniques to replace work that were not usually enjoyable for humans, and often not safe. What makes up a lot of the joy of being human is creativity, and this is the first time that the creative process itself is under siege. A prompt is a small part of the creative process, it’s an overall charter, and so many humans that are involved in the rest of the creative process are going to be made redundant. Humans doing the enjoyable parts of creative process will stop making economic sense

AI brings disruption, but also doors to new skills and creative tools. Focus on responsible innovation, not just the bumps on the road.

Sure, but are you seeing sufficient ‘responsible’ innovation? I just don’t see it. One of my points is about those tools, and that’s awesome. But I don’t see any focus on the bumps on the road.

I think this situation is exacerbated by the non-AI related layoffs that started in the tech industry two years ago or so. Companies, for the first time in my thirty year working career, started laying off en masse for near-term profit maximization rather than to mitigate perceived survival risks.

+1, and yes I think things will eventually stabilize, I am drawing attention using my network so that ‘eventually’ comes sooner than later.