Here we go. Scott is talking about AI again.

This spring, a handful of commencement speakers brought up AI while talking to graduates about their future careers. Just the word was enough to get booed. Not a specific use, not a specific harm — just the word. Yeah. That reaction tells me this issue has gotten more divisive than it needs to be, and I think a better test than “was AI involved” might help some of us make peace with it.

I don’t think most of what upsets people about AI is actually about AI. I think it’s one of two older questions, posing as an AI issue. Portrait painters said something similar about photography back when it first showed up — that anyone could do it, that it took no real skill, that it devalued what an artist actually does. I disagree. Bad photographs still got taken. (And still do.) Bad portraits still got painted. (Mostly be me.) The tool never guaranteed the artistic merit. It should not have been about how realistic the image looked. It was whether it told the truth, and whose eye was behind it.

Question one: is this telling the truth?

Some AI uses fail because they misrepresent something real to an audience trusting you not to. A destination marketing client hires us because a place is real and worth visiting. The moment we generate an image of a beach that isn’t quite that beach, we’ve broken the one promise the category depends on. Same with swapping in an AI stand-in for principal talent. The audience thinks they’re watching a person, and they’re not.
None of this is really about AI being bad. A staged candid photo has the same problem. A body double with no disclosure has the same problem. The offense is misleading someone about what’s real, in a category where what’s real is the entire pitch.

Question two: whose judgment is actually in this?

This one’s more personal. There’s a creative on my team — let’s call her “Jackie” — who said that using AI for concepting makes it feel like the ideas are no longer hers. I get it, but I think we can be more precise than “AI touched it, therefore it‘s not mine.” Didn’t we all used to pore through the awards annuals looking for ideas? (I know it wasn’t just me.) We have always looked wherever we could to send our brains into new directions. I think an AI creative partner is easy to defend as long as your judgment steers the result.
When I work with my AI platform of choice — let’s call it “Claude”— it’s a back-and-forth session. I bring a nugget, we see where it goes, I reject most of what comes back. But there’s usually something to run with or at least a different way to ask the question. The eventual output has my fingerprints on the decisions even when the words came from the model. That doesn’t feel like giving up authorship to me. It feels like arguing with someone smart until we land on something better than either of us started with.
What actually erodes authorship is using a response as-is, unexamined. Or asking AI to generate the idea itself instead of starting from some germ of thought you have. That’s where slop comes from. You could do the same damage by passing along your own first thoughts without taking the time to weed out the mediocre ideas and tighten up the good ones.

Where the two questions cross

We just produced a TV campaign for Visit Sanford where the closing montage in each spot is about 80% AI-generated clips — built from our own original photography of the destination — cut together with the remaining 20% shot on location. Run that through both tests: nothing in it misrepresents Sanford, since every clip traces back to a real photograph of a real place, so it passes the truth test cleanly. It passes the authorship test too, because the choice of which images to animate, how, and where to cut to real footage was ours, made on purpose.
Compare that to a few other cases:

• A generated image of a place that doesn’t exist — fails the truth test outright.
• Shortening a block of copy — passes both. It’s editing, not origination.
• An AI draft used word-for-word, unreviewed — passes the truth test, fails the authorship test.

Four cases, four different verdicts. You could also argue that a third test is needed: is it good? We’ll save that for another discussion.

What I can’t promise

I don’t think there’s a single clean line we shouldn’t cross, and I’m suspicious of anyone who claims to have found one. I expect my own views on this are a work in progress. Culture hasn’t settled on this yet, the same way acceptable photo retouching has moved more than once in my lifetime.

So instead of a rule, two questions I ask before AI-enhanced work goes out the door:

• Am I telling the audience something is true that isn’t?
• Could I explain, honestly, what my own judgment contributed to this?

So here’s what I’d actually say to those graduates who booed at the mention of AI: I don’t think you’re wrong to be wary. But you need to aim the boo more precisely — at the fabricated place, the unexamined draft, the idea nobody actually thought through — instead of total dismissal of AI.