The AI Factor

The AI hype machine sells a simple promise: push button, get brilliance. Fast.

The uncomfortable part is that the button works, at least some of the time - studies split on whether AI work outperforms human work. System1(opens in new tab) tested gen-AI ads and found them scoring above average on emotion. Ipsos tested them against human-made work and found the humans ahead. The headlines diverge, but the fundamental finding in both studies was the same: AI work was less original and less memorable.

Why? Because AI loves mediocrity. By nature, it regresses to the mean. What it delivers, reliably and at scale, is competent. Competent is now available to everyone, which means it becomes undifferentiated and cheap.

That leaves the outlier — the idea that takes a real strategy, a few wrong turns and an unbridled fear of the average. It is also the least forecastable thing an agency does, which is why it keeps losing the room to the option with a number attached.

The mean is not an AI problem. It is where everyone defaults.

The work that escapes it comes from friction: agency teams willing to argue with each other and with the machine, instead of accepting the first answer. The machine can generate more options than any room of humans could. It cannot tell you which one is brave or brilliant. That is where human judgement, which agencies have relied on for a century, returns. The machine does not replace the judgment. It expands the field you judge.