For the past few days, ChatGPT, the artificial intelligence developed by OpenAI, has been making a lot of noise. This tool uses an advanced language model to generate written content on its own. Capable of writing articles, jokes, poems, and computer programs, he would also be able to make search engines as we know them obsolete.
Will Google be killed by ChatGPT?
This is what Paul Buchheit explained on Twitter. The creator of Gmail, who left Google in 2006, explains: “Google may only be a year or two away from total upheaval. The AI will eliminate the search engine results page, where they make the most money. Even if they catch up on AI, they can’t fully deploy it without destroying the most valuable part of their business! »
He compares the fate of Google to that of the Yellow Pages, which have become obsolete over time: “It was also a phenomenal business model, until Google came along. AI will now do the same for search engines. »
This vision is not unanimous in the world of Tech. As reported 20 minutes, Nicolas Weaver, researcher in computer security and networks at the University of Berkeley, is not at all of this opinion. Indeed, as he points out, ChatGPT does not source its answers, which does not guarantee their reliability. He adds that Google already integrates AI “for years, but with large amounts of text and not just one-word-at-a-time answers”.
Google is working on its own text-generating AI
In fact, we were talking to you last year about this internal research program at Google. The idea is to use increasingly advanced language models to allow users to ask a question, while the tool is able to reliably answer it. Google could then cross several sources to give satisfaction to the Internet user, as a human does when he synthesizes information in writing, in particular on Wikipedia. You can also read this working document here.
Very impressive, ChatGPT nevertheless retains many shortcomings. As we explained to you this week, we had this AI write a short article. If her French is correct, her database stops at the end of 2021. She also does not master the notions of angles or information verification.
For your part, have you tested this artificial intelligence, and if so, what did you think of it? Tell us in the comments.
