Helping people to discover the content of the web remains central to our approach, and with IA mode, we facilitate the exploration and action of people. With the search for deep information of the model, people can better express what they are looking for – with all their shades and constraints – and go to the right web content in a range of formats.
Laboratory test
We have obtained internal comments and trusted testers, and they found the IA mode incredibly useful – they particularly appreciate the speed, quality and freshness of responses.
Now we expand our tests with a limited opt-in Laboratory experience. This experimental approach helps us to learn what is most useful and to improve quickly with the comments of people most wishing to try it.
AI mode is rooted in our nucleus Quality and classification systemsAnd we also use new approaches with the model’s reasoning capacities to improve billing. We aim to show as much as possible an answer fueled by AI, but in cases where we do not have high confidence in utility and quality, the answer will be a set of web search results. As with any AI product at an early stage, we will not always have things correctly. For example, although we aim for AI’s responses in research to present objectively information according to what is available on the web, it is possible that certain answers seem involuntarily taking a personality or reflecting a particular opinion.
In this next test phase, we will take up these types of challenges and will also quickly bring the user experience according to the comments we obtain. We are already working on new capacities and updates, such as adding visual responses with images and videos, richer formatting, new ways of obtaining useful and much more web content.
From today, we will start to invite Google One Ai Premium Subscribers are the first to try this experience in laboratories. We look forward to the comments and are listening to find out more!