Search Is Learning To Understand The Real Question

Fox put it simply: “The biggest thing that AI is enabling in search is it’s enabling people to ask questions they could never even ask before.”

That is a powerful idea. In the old world of search, users often had to translate a real need into a few clunky keywords. Search worked, but it forced people to think like machines. The new ambition is the reverse. Machines should do more of the work so people can ask what they actually mean.

Fox gave a good example. Instead of searching for something vague like “memory card,” he asked for the right memory card for a specific router and a specific purpose. AI Mode could interpret the context, understand the technical requirements, and return a much more useful answer.

This gets to the heart of where search is going. The future of search is less about hunting through links and more about getting to the best answer faster, with enough context to make that answer genuinely useful.

That does not mean Google is trying to turn Search into a pure chatbot. In fact, Fox said, “A user shouldn’t need to choose between a chatbot and a search engine.” That is a crucial point. Google’s strategy appears to be blending reasoning, summarization, and web results into one experience, instead of forcing users into separate modes.

From my perspective, this is the most commercially important AI product battle happening today. Billions of people use search as a gateway to the internet, to commerce, to media, and to daily decision-making. If Google can reinvent that gateway without breaking it, it will have achieved something extraordinary.

Trust, Ads And The Future Of The Web

One of the biggest questions hanging over AI search is what happens to the open web. If AI gives us answers directly, do we still click through to content? And if fewer people visit websites, what happens to publishers, creators, and the incentive to produce quality information?

Fox is strikingly optimistic here. “I consider myself a web optimist,” he told me. He believes people still want to go deeper, especially when they care about a topic. AI summaries may help people get oriented, but many users will still want richer content, deeper analysis and original reporting.

I think he is right, up to a point. The web is unlikely to disappear. What will change is the role that different kinds of content play. Commodity content is under pressure. Thin content built mainly to attract clicks is likely to struggle. Strong content with expertise, personality or genuine depth has a much better chance of thriving.

Advertising, of course, is another big part of this equation. Google has built one of the most successful ad businesses in history, largely on the back of search intent. As search becomes more conversational, that model has to adapt. Fox was very clear that ads will remain “clearly marked, clearly indicated” and separate from organic responses. He also made the important point that ads work best when they are relevant.

That sounds obvious, yet it is essential. In the AI era, users will not tolerate experiences that feel manipulative or muddy. Trust is fragile, especially when AI is generating answers in a confident voice. If Google gets this balance right, ads could become more useful and better timed. If it gets it wrong, user trust could erode fast.

When Search Starts To Act

Beyond answering questions, AI is starting to do things for us. This is where the discussion becomes even more interesting.

On the question of agentic capabilities in search, Fox was both enthusiastic and thoughtful. He described a feature launched around six months ago, now rolling out internationally, that aggregates availability across restaurant booking platforms like OpenTable and Yelp, so a single query can surface real-time reservation options across all of them simultaneously. He used it himself the same week we spoke, finding a last-minute Friday table at a sushi restaurant he would never have thought to check.

This is an early glimpse of what the next phase of digital assistance will look like. We are moving from systems that know things to systems that do things.

That said, Fox also made a point I strongly agree with. The ceiling is not defined only by what the technology can do. “The ceiling that matters the most is what do users want?” he said.

That is exactly right. People want convenience, yet they also want agency. They may be happy for an AI to compare booking options or narrow a list. They may be less happy for the AI to spend money, choose a vacation or make more personal decisions without their input. "I personally don't believe in a world where everything is agentic because I think people like personal agency," he said. The winners in agentic AI will be the companies that understand this balance and design for it carefully.

Maps Becomes A Much Smarter Guide

Search is not the only Google product being reinvented. Maps is going through its own AI transformation.

Fox told me that Google is using AI to process imagery from Street View cars, aerial sources, and other inputs in order to “actually build the map.” That may sound technical, yet it is hugely important. It means AI is helping Google keep the physical world model up to date and usable, identifying changes such as traffic lights or businesses that have closed.

On top of that, Google is making Maps more conversational. With Ask Maps, users can ask far more specific questions, such as where to find a family-friendly restaurant that serves shrimp tacos nearby that are open at the right time. That sounds like a small improvement. It is actually a big one. It shifts Maps from being mainly a navigation tool into a decision support system for life in the physical world.

I suspect this is where some of the most practical consumer AI value will emerge over the next few years. Many of our daily decisions are local, contextual and time-sensitive. A smarter Maps experience has enormous potential.

Personal Intelligence And The Next Interface

Perhaps the most forward-looking thread in our conversation was about personal intelligence, the idea that Google can make search dramatically more useful by drawing on information users have already entrusted to it across Gmail, Calendar, Photos and Maps.

Fox's example was vivid. While skiing in low-light conditions, he asked AI mode what ski goggles lenses would work best. Personal intelligence surfaced an email his wife had sent him years earlier, the receipt for a pair of goggles she had gifted him, complete with details about the two lenses that came with them. The system then reasoned across that information and the weather forecast for the following day to recommend which lens to use. As Fox put it, the result can feel “subtly magical.”

This is where AI starts to become truly personal. And this is also where the trust challenge becomes very real. The trust argument Fox makes is that this works precisely because no data is leaving Google's ecosystem. Users opt in, the information is already securely stored, and the system is transparent about where its answers are coming from. That combination of consent, security, and transparency is his answer to the question of how you build user trust around something as sensitive as personal data in search.

Google’s Internal AI Transformation

Fox also spoke about how AI has changed Google's internal development practices. Fox, who has been at Google long enough to have lived through every major wave of innovation, described agent-based development as “a sea change,” saying, “I’ve never seen Google operating at the speed that it’s operating at right now.”

Engineers are now managing coding agents rather than writing most code directly. Product managers can build working prototypes instead of writing requirement documents. Overnight, an AI system analyzed evaluation data across dozens of language and country segments for a model update, and delivered a structured report on where performance gaps existed and what might be causing them.

This matters far beyond Google. It is a glimpse of how product development itself is changing across the tech industry. AI is becoming part of the machinery that creates the next wave of AI products. That feedback loop will accelerate innovation dramatically.

The Next Search Era Has Already Started

If there is one big takeaway from this conversation, it is that the future of search will be more conversational, more contextual and more action-oriented. The search box is staying, at least for now, yet what happens after you type into it is changing fast.

None of that comes without its challenges. Google has to make search smarter without losing trust, support the web while adapting its business model and make AI more personal without making users uncomfortable. It has to introduce agents in a way that feels helpful rather than intrusive, all while competing with the likes of ChatGPT and Claude for the attention of people who now turn to conversational AI before they ever open a browser, and with the likes of Grok and Meta AI reaching users directly inside the social apps they already live in. There has never been more competition for where people get their information, and Google is fighting on every front simultaneously.

That is a difficult balancing act. Yet if Google succeeds, search will become something much more powerful than an index of the Internet. It will become a layer of intelligence that helps us navigate information, the physical world, and increasingly the tasks that fill our days.