Google removes AI feature from Search that showed medical advice from random users

Google has removed the What People Suggest feature from its search, which displayed medical advice from random users worldwide. The decision looks like a rollback amid growing pressure on the company over how AI should handle health topics.

What exactly was removed

The new search feature What People Suggest was at issue. It displayed health advice collected not from doctors or clinics, but from ordinary users. Essentially, Google tried to embed a "others' experience" format into search and present it as another useful layer on top of traditional search results. For everyday topics, such an approach can still be debated, but in medicine, the cost of error is completely different.

Previously, the company described the launch as an example of how AI can improve access to health knowledge on a global scale.

"This demonstrates the potential of AI to transform health outcomes worldwide."

The problem is that there's a huge distance between "useful experience from other people" and "advice you can trust" in medicine. When a user searches for information about symptoms, treatment, or self-diagnosis, they need not just a quick answer, but an answer with a clear level of reliability.

Why questions arose

Personal experience is not useless in itself. People often seek it specifically: how someone tolerated treatment, what helped manage side effects, what questions to ask a doctor. But when such answers are embedded directly in a major platform's search, they start to look more weighty than they actually are. In Google's interface, a stranger's private opinion is easily perceived almost as a verified recommendation.

The weak points of such an approach are immediately apparent:

• advice may be based only on personal history, not clinical data
• another user may have a different diagnosis, age, medications, or contraindications
• AI makes disparate opinions visually more cohesive and authoritative
• global collection of advice doesn't account for local medical standards and treatment availability
• even without direct error, such an answer can push toward dangerous self-diagnosis

This is precisely why everything related to health remains one of the most sensitive areas for AI companies. Here it's not enough to simply "collect what's useful from the internet" or retell user experience nicely. You need transparent frameworks: where the recommendation came from, who gave it, how verifiable it is, and where information ends and potentially risky medical advice begins.

Why this matters for Google

The decision to abandon the feature comes against growing scrutiny of how Google uses AI for health suggestions. The fact of such a move itself shows that even the largest platforms haven't yet found a safe product formula for such scenarios. In ordinary search, an error is unpleasant. In medical results, it can cost a user time, money, and in the worst case, their health.

For Google, it's also a question of trust in search itself as a basic service. The company has spent years training its audience that search is the entry point to the most important information. When an AI layer with advice from random people appears within this experience, the boundary between reference data, community opinion, and almost a medical recommendation starts to blur. If this boundary can't be explained in a few seconds, the product becomes too risky for a mass audience.

What this means

The What People Suggest story shows a simple limit for AI search: the more sensitive the topic, the less the market is willing to tolerate an experimental approach. In news, shopping, or travel, an error is annoying; in health, it quickly becomes a reputational and ethical risk. For the entire industry, this is a signal: AI can help search for medical information, but replacing verified expertise with crowd advice is premature.