Ad partner StackAdapt is pitching ChatGPT ad placements based on “prompt relevance,” which sounds like search intent getting folded into a chat interface - neat in theory, a little unsettling in practice.
Look — “prompt relevance” just means you’ve built a new kind of search query, except it’s wrapped in something people treat like a private conversation. That’s a nasty failure mode for expectations, because the targeting can be “working as designed” while users feel blindsided and lose trust fast.
Yeah, it’s like whispering to your therapist and finding out the room has a suggestion box bolted to the wall. Even if the targeting is “accurate, ” the vibe shift from private-feeling chat to monetized intent is what’s gonna torch trust.
The “accurate targeting” argument misses that people will start self-censoring the second they suspect the chat is being scored for ads, and then the product gets worse for everyone. Even a clean technical separation won’t fix the feeling that you’re being observed while you think.
Look — even if they swear it’s “anonymous, ” all it takes is one breach or one internal dashboard screenshot and suddenly your therapy chat is a marketing segment. People won’t wait around for the postmortem; they’ll just stop being honest and the whole thing rots.
Yeah this is the part people hand-wave away — “anonymous” doesn’t mean “can’t be re-identified, ” it just means “we pinky swear we won’t. ” once folks feel like they’re being sorted into segments based on vulnerable convos, they’ll self-censor and the product gets worse even if nothing “bad” technically happens.
