On Advertising Inside AI

ChatGPT has over a 100 million users in India, and Google is ensuring that its AI models are in every surface that Google owns. The answer to the questions about where the worlds advertising will go, even as the “answers” engine destroys and replaces the search engine is obvious: that money will have to find somewhere to go, and hence it will land up in AI and Social. A few points to consider:

On the consumer front:

1. In India, ads are fine if AI Is free: most users won’t object to advertising if the product becomes free or cheaper. That’s the tradeoff users in India are used to making. There won’t be a backlash about ads inside AI, except from those folks who tend to forget that advertising is the primary force behind making the web free. I’m waiting to see how will deal with the ad blocker problem.
2. Advertising in AI isn’t the same as search: Unlike search, AI chat environments feel immersive and almost confessional, because either users disclose anxieties about money, health, relationships, and career decisions, or it can be inferred from the conversation. Anthropic got its superbowl advertising right: it is going to feel horribly intrusive as ads enter the conversation. Ads placed adjacent to chat (maybe even a 300×250 or a 160×600 on desktop or an interstitial on mobile) will probably just become blind spots. At the same time, an ad might just pull a user out of immersion by making them conscious about what they’re sharing. AI companies will have to be careful.
3. Separation between answer and ad becomes ethically critical: For those without significant critical thinking skills, in an immersive environment where the user is vulnerable, the gap between sponsored content and an endorsement narrows, to the point where it appears like a genuine suggestion. AI tools will have to have clear markings (not tiny light gray text that says “Ad), and the norms for that haven’t been set yet.
4. Ads in regulated categories: Search ads respond to explicit intent (“best car loan rates India”), while Chat can be used for answers and advise: “Should I refinance my home loan?”. Advertising becomes tricky in regulated domains like health, trading, accounting and lending. In India, regulators such as RBI and SEBI distinguish strictly between advertisement and advice. In search, doctors relied on SEO and advise, and it was all right if it came directly from them, unpaid. It’s third party endorsement that becomes an issue: I’m wondering if we’ll see regulators looking more closely at how apparent automated endorsement, even if not paid, shapes selection in this environment. 
5. Two questions on liability, again related to the narrowing gap between endorsement and advertisement:
   • At what point does a suggestion feel like a recommendation?
   • Who is liable for recommending a bad product?
6. Will inference finally become a privacy consideration? Google so far had intent based information, and demographic…