An AI jury picked a different winner
Ad Age trained an AI jury on past Cannes winners and ran it against the 2026 Titanium shortlist. It picked a different Grand Prix than the human jury. That gap is the limit of the machine.
Mariano De Vitto · June 2026
Cannes Lions opened this week with its usual flood of metal, and one experiment cut through it. Ad Age built an AI jury, trained on past Cannes winners, and ran it against the 2026 Titanium shortlist. The machine picked a different Grand Prix than the human jury did. The headline is fun, the implication is the story.
Titanium rewards work that breaks the rules, judged this year by a panel led by Chaka Sobhani of TBWA Worldwide. A model trained on what won before recognizes the patterns of past greatness, but cannot feel the jolt of something never done, which is what Titanium exists to reward. The gap showed across the festival: AB InBev won the inaugural Creative Brand Lion, tied to business results, and The Ordinary took Health and Wellness for attacking the beauty industry pseudo-science language.
What it tells marketers
This is not an argument against AI in creative work. The same week, every platform shipped generative tools, and they are useful. It is an argument about order: the model is excellent at scoring, sorting and generating variations, and weak at deciding what deserves to exist, because that depends on context and taste. The AI jury rewarded the safe echo of past winners. The human jury rewarded the leap.
The signal: A model can grade creative against the past. It cannot choose the leap. Use it to widen options, not to pick the winner.
The Signal Brief · Mariano De Vitto — Head of Marketing, Barcelona