The second are tools for advertisers. If you run a petite business, you don’t think about all the queries people will be typing in. Artificial intelligence is great at figuring out what keywords to operate, what the optimal innovative is, and generating it all.
The third element is the most nascent: advertising in modern experiences. The general philosophy we have is to build a great consumer product and then think about monetization. Because the business is so mighty and robust, we have the luxury.
What have you learned from your AI advertising experiments?
Ads are always separated from organic results and clearly marked. If we determine that no ad is relevant, we will not show any ads. Perhaps the most crucial rule is that ads should be useful.
What [ads in AI Mode] have shown that these are mostly intuitive things. If it’s relevant, the user will click on it. If not, they won’t do it.
In Davos, Demis Hassabis said Google had no plans to share Gemini ads. How do you think about it now?
The reason we focused on AI ads and AI Reviews is because we see them as extensions of search. This is the most natural place for us to conduct initial experiments.
I would expect that the lessons we learn from AI advertising will likely carry over to what we might want to do in the Gemini app in the future. We’re able to get all of these insights in a context and design that users already know the ads are familiar with. It may seem strange, but our research shows that users actually like ads in the context of search. Over time, we’ll figure out what makes sense for the Gemini app.
So you’re not completely ruling out advertising on Gemini?
No, we do not exclude them. It just wasn’t our focus.
Gemini is a huge product right now – it’s growing quite quickly. OpenAI is already trying out ads in ChatGPT. Why do you think Gemini is not ready for advertising?
We are very pleased with how well it is growing. I would say it’s more of a matter of prioritization – what should you focus on now?
Do you think OpenAI introduced advertising too early?
It’s challenging for me to say. I think it will really depend on how they do it. We have seen that creating ads correctly increases the user experience of the product. But what’s really crucial is to get it right, and that means relevance, quality, and not placing ads where users don’t want to see them.
It’s challenging to do. We have over 20 years of experience, we learn how to do it. It’s less a matter of time and more a matter of doing it right and in a way that respects users.
Recently, Google launched Personal Intelligence feature in Gemini and AI mode. I imagine advertisers would love to get their hands on this data. How do you think about it?
Personal intelligence is extremely useful. For example, I was skiing and I couldn’t see well through my glasses – it was gloomy. I asked the AI mode a rather vague question about what lens I needed for which conditions. From my email I was able to find out what resort I would be staying at, what mountain I would be skiing and what the weather would be like. He also found the receipt that my wife sent me in exchange for the goggles – it turned out that she had bought me an additional lens. It’s like subtle magic.