Should Google fear OpenAi's entry into the search market?

The comparison with Gemini

Another fact of reality, this time technological: Gemini, which so far has not impressed like Gpt4 - it must be said - has a function that the others do not have. Gemini, which also has access to the latest information found on the web, offers the possibility of verifying the response produced by the chatbot. Basically, it searches the web for information that can confirm the generated text. The sentences in green contain verified information that Gemini has found a match for on the web, those in orange, on the other hand, are those that are likely to differ from the information retrieved on the web. It is not a litmus test of the veracity of the information expressed by the chabtbot also because even in the case of a green light Gemini does not guarantee that the sentence was generated from that link. But it is an aid to reducing hallucinations. It must be said that to date and for all existing chatbots, no lie detectors have yet been invented.

The Google-Apple relationship

Then there is financial data. Let us take the case of Apple, which as we know is the second largest smartphone manufacturer in the world. Google pays Apple 20 billion every year to be the defaul search engine of every iPhone and Safari browser. It's hard to imagine that both of them don't pay each other.

How will searching with Ai be?

The fourth fact is related to the functioning of the engines with the ai. Cas we explained to date, they basically summarise search results, offer citations for its answers, and help users refine their questions to get the best answers. For information seekers, the change is substantial because you go from a list of links to click on to someone who answers your questions and tries to understand what you want to know. As you use these tools a bit for work and study, you find that they work better, especially if the questions are complex. If, for instance, you are not clear yourself what you are looking for, if you are moving in terrain you do not fully master, if you need questions to your questions, the possibility then to share what you are asking with others makes research a shared task with prompts from several people together. If, on the other hand, you need more precise, simpler and more immediate information, such as the age of the actor, the result of the football match or the location with certain characteristics, Google Search, to name but one, remains the fastest, most precise and most efficient choice today. So if you trust it, and use it as a traditional search engine, it can generate misinformation because a natural language response is by nature more assertive and convincing than a list of links from which it is easier to distance oneself.

So for whom is this a problem?

Summing up, as a good chatbot would, OpenAi will not challenge Google, which tomorrow during the annual Google I/0 developer conference will show new integrations of its most powerful LLM Google Pro 1.5 into its ecosystem. It will, however, stimulate it to quickly change its business model for Search Generative Experience (SGE), which is the application of chatbots to search engines. . The impression is that Google has everything very clear from a technological point of view on how to make the search engine work with AI. It perhaps lacks a clearer idea about the business model, i.e. how to make money from these new tools.

It certainly makes life more difficult for Microsoft, which will have a harder time differentiating to its shareholders the difference between Bing and OpenAi's new creature.