Thought Leadership
Chatbots such as ChatGPT have often overshadowed Google recently – but the internet giant is now going on the counterattack.
At the Google I/O developer conference, the company presented its vision of how its artificial intelligence should accompany users at every turn in their everyday lives and make their lives easier. Cameras on smartphones and computer glasses play a central role in this, allowing AI to see what people are looking at.
The heart of the plan is called Gemini – Google’s answer to ChatGPT and Co. Under the hood of Gemini is a jumble of AI models for various tasks ranging from personalized web searches to automated purchasing advice.
“Universal AI system”
At I/O, Google managers left no doubt about the company’s ambitions. “The ultimate vision for the Gemini app is to turn it into a universal AI system,” said Demis Hassabis, the top manager responsible for artificial intelligence. This AI system should be tailored to users and offer them its services proactively instead of just reacting to their requests.
The basis for Google’s advance is its role as the dominant search engine for years, which captures virtually all information on the web and has a direct line to users. In competition with its new AI rivals, Google is also improving its web search with artificial intelligence. The vision is that instead of a list of web links, users will receive more detailed answers and be able to ask further questions.
Google took the first step in this direction a year ago with the AI summary of information above the other search results. This function, which is called “Overview with AI” in Germany, already reaches more than 1.5 billion users, said CEO Sundar Pichai.
Next level for search queries: AI mode
The next stage is the AI mode, which can provide more detailed and reasoned answers to search queries. For example, the AI mode could learn to provide purchasing advice when searching for a household appliance.
If you are looking for an apartment or tickets for an event, the software should be able to search the web independently to find the best selection. If desired, the program could also book a viewing appointment for an apartment in this so-called “agent mode”. If you have set up a price alert for the product search, Gemini should also be able to strike.
Instructions for bicycle repair
In another demo, Google showed how AI can help repair a bicycle. Not only can you ask Gemini for suitable spare parts and ask it to get them, you can also point the smartphone camera at individual screws and ask it what tools you need.
To keep hands free, the camera in computer glasses will take over this role. At I/O, Google also showed prototypes of slim glasses that can display information in the user’s field of vision. However, it remains to be seen when these devices could come onto the market.
ChatGPT and co as challengers
Google has dominated internet searches for more than two decades. However, users are now also turning to AI chatbots such as ChatGPT or Perplexity. While Google still primarily presents a list of web links as a result, at least in part, the new rivals are concentrating on providing direct answers.
It is unclear whether Google is already suffering as a result. The company earns its money primarily from advertising in the web search environment – and the business has continued to grow recently. At the same time, Apple’s top manager Eddy Cue recently caused a minor earthquake when he testified in court that Google queries from the iPhone company’s Safari web browser had recently declined for the first time. Google countered that it was still seeing an overall increase from Apple devices.
Even in AI mode, links from advertisers could be seen at the edge of the screen in exchange for money – after all, advertising revenue must continue to flow. This made industry analyst Gene Munster skeptical: “I don’t think a simple answer wrapped in advertising is what consumers want.”
AI learns about users from emails
Users should be able to personalize the search more by giving the AI access to their Google apps such as Gmail. If you are traveling, the software would then recommend a visit to an art gallery, for example, because it would know that you are interested in it thanks to newsletters you have subscribed to. It could also know from previous search queries, for example, that people prefer to sit outside in restaurants.
Some new functions will initially only be released in the USA or in the experimental Google Labs area, which is not yet available in Germany. Others will initially be offered in the Google AI Ultra subscription, which will initially be available in the USA at a price of around 250 dollars per month.
AI is also the focus of other Google announcements at I/O. For example, the company is entering the virtual fitting room business. The software attempts to use knowledge about the user’s physique and the material of the clothing to calculate how the items will fit as well as possible. With Veo 3, Google is improving its AI software that generates videos so that it is also suitable for professional use. For the first time, the model can also automatically generate the right sounds for the image.
dpa
