That’s fair, and I think your point about the pipeline shrinking from the bottom up is a real concern.
On the coding side, I’ve been using AI myself without any formal background, and it’s definitely powerful. But what I’ve found is that it still depends heavily on the person guiding it.
I can usually get something working because I understand the workflow and what the end result should look like. But getting something efficient or well-structured is a different story. I’ve had cases where an experienced developer reduced a 100-line script down to 10 lines, and when I tried to get AI to replicate that approach, it struggled even with a working example.
So it’s not that AI can’t do it, it’s more that it still relies on someone who understands the problem properly, not just someone prompting at a high level.
I’ve seen a similar pattern in prepress over the years. When I started, there were multiple roles across the department, typesetters, designers, proofing, imposition, plates, management. Within a few years, that was reduced to just a couple of people as the technology improved.
So I agree that AI will likely continue that trend, doing more with fewer people.
Where I’d differ slightly is that I don’t think it removes the need for people entirely. It reduces the numbers, but the people who remain need to be more capable and more adaptable. Even getting consistent results out of AI tools requires a level of understanding that not everyone will have.
So it feels less like replacement, and more like compression, fewer roles, but higher expectations on the ones that remain.
That’s a real concern.
