In the first episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about two opposite moves happening in AI and marketing: OpenAI putting ads directly inside ChatGPT answers, and brands loudly promising they do not use AI at all.
The episode starts with what ChatGPT ads actually look like. They are not banners off to the side. They are sponsored cards that appear underneath an answer, inside the same conversation users already trust. Harjot sees a major opportunity: people tell ChatGPT things they would never type into Google, creating a new kind of intent signal for advertisers. Will is more cautious. The signal may be rich, but the controls are still thin, the measurement is limited, and the ad can feel uncomfortably close to a recommendation.
The second half turns to the anti-AI backlash. Aerie, Equinox, McDonald’s, and Anthropic all become examples of brands trying to work out what “human-made” actually means. Harjot argues that the enemy is not AI, but slop. Will argues that “no AI” can work if it is rooted in a real brand history, as Aerie’s is, but becomes dangerous when it turns into a promise the company may not be able to keep.
Highlights of this Episode Include:
- ChatGPT Ads Are Not Banners: Sponsored placements appear inside the trusted experience, underneath the answer, which makes them feel materially different from traditional display ads.
- Intent Plus Trust Is the Bull Case: Harjot argues that ChatGPT captures unusually rich context because users explain what they want in full, not just through keywords.
- The Controls Are Still Thin: Will pushes back on the current product: limited targeting, no retargeting, weak measurement, and no clear way to know whether early spend creates durable advantage.
- Before You Buy the Ad, Check You’re in the Answer: Will suggests turning ChatGPT memory off and asking it to recommend products or services in your category before spending money on sponsored placement.
- The Anti-AI Backlash Is Really an Anti-Slop Backlash: The episode argues that bad AI work is not bad because AI was involved. It is bad because there was no taste or judgment in the process.
