Google is improving its search engine and other services with artificial intelligence. Overviews of search results generated by AI will soon be available to all users in the USA and other countries will follow, announced CEO Sundar Pichai at the start of the Google I/O developer conference.

In competition with the popular chatbot ChatGPT and other software with artificial intelligence, Google relies on its in-house AI model Gemini. The new motto for web searches is “we do the Googling for you,” said the responsible top manager Liz Reid. You can now ask several questions in one search query – for example, not only where Pilates studios are in Boston, but also how to get there and book a time slot.

“Dimension of human curiosity”

Several AI startups are trying to give Google more competition in internet searches. They rely on providing users with concrete answers more quickly instead of web links. If such search behavior becomes commonplace, it could jeopardize Google’s current business model, as the company still makes the majority of its revenue from links from advertisers that are placed next to search results.

So far, AI competitors such as Perplexity AI have not been able to seriously challenge the Internet giant’s dominance in the business. Google itself has already introduced, among other things, a function in which all it takes to search the Internet is to circle an object in a photo or words on the smartphone display.

With the new functions, the group itself is going further beyond the concept of useful links. “Google Search is generative AI at the scale of human curiosity,” said Pichai. Another example was that it should be enough to feed the search with a picture of a defective record player so that the AI ​​software can find solutions to the defect.

The new AI-powered search will initially be launched in English in the US. But it should come to Europe and Germany “in the foreseeable future”. “We have been investing in AI for more than a decade,” emphasized Pichai. And yet the technology is only at the beginning.

Evaluate 1500 pages at once

Among other things, Google is working on expanding the so-called “context window” – the amount of information that an AI model can evaluate at the same time. The subscription version of the AI ​​software can currently capture a PDF document up to 1,500 pages long or a one-hour video at once and answer questions about it. No other chatbot is capable of this, emphasized Google manager Sissie Hsiao. They want to double the values ​​at the end of the year.

Google is also relying on “AI agents” – assistants that can independently carry out multi-step tasks. You could, for example, take care of returns completely or, if you move, do the necessary re-registration and find useful addresses in the new neighborhood.

Among other things, Pichai demonstrated how the software can now select all images from the collection of personal photos that, for example, show a child’s swimming progress. Finally, the Google boss asked Gemini to count how often the term AI was used in the almost two-hour presentation: it was more than 120 times.

The competition is showing off

The day before, ChatGPT developer OpenAI had set the bar high for Google’s announcements. OpenAI presented a version of ChatGPT that can converse fluently with people and also detect their state of mind.

Google had been experimenting with AI software for years, but the chatbot ChatGPT overtook the Internet giant’s software when it entered the market. Google first demonstrated in a video a version of its AI called “Project Astra” that allows you to chat without delay and ask questions about objects it sees through the smartphone camera.