Eric Hoffer once wrote that “Every cause begins as a movement, becomes a business, and eventually degenerates into a racket.” AI has moved through those stages fast, shifting from a breakthrough technology into a commercial interface.
This transition became harder to ignore when Perplexity dropped advertising over concerns it could damage trust in AI. The decision matters beyond one platform, because it points to something the industry still has not properly resolved. Once commercial influence is folded into a conversational response, it becomes much harder to tell where useful guidance ends, and a sales message begins.
That is what makes this advertising conundrum feel different. AI assistants are not being used like old search engines. People have moved beyond typing in quick factual questions or looking for a route to a website, and instead are asking anything from what to buy, to what hiking boots are best for the Yorkshire Dales. In plenty of cases, they are using these tools in the same loose, casual way they might once have used a group chat or a late-night Google. In some cases, they are also using them more privately than that, as a place to unburden worries, test half-formed thoughts and ask the kinds of questions people might once have kept to themselves.
Search previously always came with more distance. Even if most people were not analysing the mechanics too closely, the value exchange was still there in plain sight. You were shown options and left to make something of them yourself. AI instead shrinks all of that into one answer, delivered in a tone that feels tidy, direct and self-assured. That is part of the allure. It can sound as if the hard work has already been done for you.
When the answer starts sounding like the sales pitch
Advertising in search is easy enough to grasp. Advertising inside a conversational response is much murkier. The message does not sit neatly beside the guidance. It can sit inside it, folded into the same tone and apparent authority. Instead of weighing up a few sources and deciding what feels credible, the user is handed a response that already sounds settled.
That changes the experience in a bigger way than the industry sometimes admits. If Google, ChatGPT, Gemini, Claude, Perplexity and others become the place people go for recommendations and everyday judgement calls, then monetisation is not some background detail. It shapes the answer itself and how much faith people put in it. A sponsored listing on a search page is one thing. A monetised answer delivered in the voice of a helpful assistant feels different. It feels closer, less visible and more easily mistaken for neutral guidance.
Why a more personal interface creates a bigger trust problem
This becomes more sensitive when you think about the kinds of things people are now asking AI. Some queries are obviously commercial. Someone looking for the best trail shoes is already on a product journey, so a branded nudge may not feel especially surprising.
People are also using AI for more personal questions. They are asking about money, health, work problems, relationship doubts and all the small but loaded decisions that sit somewhere between information and advice. The sort of things they might not post publicly, but will happily type into an assistant that feels private, responsive and non-judgemental.
That intimacy is a big part of why these tools have become so sticky. They do not just return information. They create the feeling of being helped along. That is also why the commercial layer feels more fraught here than it did in earlier digital environments.
If a system starts to feel as though it is commercialising moments of uncertainty, or quietly steering people when they think they are getting a clean answer, the relationship changes quickly. What looked useful starts to feel opportunistic.
The business model may decide who people keep believing
The awkward truth is that AI is expensive to run. Free access at scale was never going to remain simple forever, and the pressure to monetise was always going to arrive. The real question is what kind of bargain people are willing to accept once platforms start trying to cover those costs more aggressively.
Paid tiers make sense to people. Usage caps make sense too. Even off-peak access models feel easier to understand than a response that sounds neutral while quietly carrying a commercial agenda. None of these options are perfect. Some will feel clunky. Some will be unpopular. But they are still clearer than abruptly introducing adverts overnight.
The platforms that hold people’s trust will be the ones that find a commercial model that does not leave users second-guessing the answer they have just been given.
Abbey Grocott, is head of content strategy at Brave Bison
