Search on the Internet had practically not evolved for 25 years, sealed by the domination of Google, which has never ceased to iterate on its initial objective set in 1998: “Organize all the world’s information and make it universally approachable and useful.”
Since then, the Mountain View company has continued to invest and innovate in this perspective, developing on a planetary scale a series of algorithms among the most complex ever implemented, optimized by thousands of engineers performing some 300,000 tests each year giving rise to 4,000 technical modifications to make the answers more relevant, all in 149 languages. Incidentally, the company has become the world’s leading advertising machine, capturing a third of global investments in digital advertising (26% today), with an annual turnover of 283 billion dollars for 2022 and a profit of 60 billion. For its part, Microsoft achieved last year a turnover of 198 billion and a net profit of 73 billion. By comparison, TotalEnergies made 19 billion in profits in 2022.
But above all, Google’s technological superiority has enabled it to capture a market share of more than 90% of global searches on the web, leaving only a measly 4% to Bing, Microsoft’s search engine. This encourages neither modesty nor questioning…
Sudden awakening on February 8
It is this supremacy that is now being brutally called into question by the emergence of so-called “conversational” interfaces. The best-known form is ChatGPT, which Microsoft has taken over, transforming it into a war machine whose sole objective — expressed in these terms by its CEO Satya Nadella — is to unbolt the statue of absolute commander of the search that Google has erected.
The offensive took the form of a presentation made at Microsoft headquarters in Seattle on February 8, barely twelve hours after a similar event organized by Google in Paris to promote its latest innovations in search. The highlight of the Parisian show was to be the presentation of Bard, a conversational agent likely to respond to ChatGPT, as well as the introduction of a concept called “Multisearch” where research is carried out through the smartphone thanks to augmented reality .
There parisian presentation from Google was initially considered a little sluggish – lack of focus, too few new features, too anecdotal use cases, bugs. Then she appeared downright corny in the face of the string presentation made in Redmond at Microsoft, which reflected rigorous mastery and a clear vision. This was embodied by its CEO, a relaxed Satya Nadella, multiplying interviews with influential media, wall street journal To The Vergeand confidently detailing his plan to challenge Google’s supremacy.
Google also pays its little regard for communication. Although the company has always known how to recruit the best in communication or public affairs to spread its message, the importance of this message has never been at the center of the concerns of its founders: “Larry Page and Sergey Brin have always been so convinced of the moral superiority of the company (“self-righteousness”), that they restrained those who were responsible for building and disseminating the message”, summed up a Mountain View executive a few months ago. Like others, he was sorry that the company was depriving itself of doing to assert its positions, to defend its fundamentals against its detractors.
The trap
Today, a revision of the doctrine is essential. Many wonder if the demand for change will rise all the way to the top, with a sympathetic but notoriously indecisive Sundar Pichai, CEO since 2015, who some Silicon Valley Kremlinologists say is ripe for a replacement.
In any case, Wall Street harshly sanctioned the performance gap between the rigorists of Redmond and the cool attitude Californian: over the last five trading sessions, Google’s share price has lost 7% while Microsoft’s share has gained more than 2%; and if we take the last 12 months, in the context of a Nasdaq which lost 15%, Google collapsed 30% against an 11% loss for Microsoft.
For those who knew all of these companies in their early days, Google rested on its laurels for too long, made no noticeable changes to its interfaces, and ended up being overwhelmed by competition it didn’t have. seen coming. Worse still: in a curious irony of technological history, Google has favored competition in two ways. The first was to release certain key algorithms into the public domain with the idea of creating a technological standard from which, through its mastery, Google would have been the first beneficiary – a classic Silicon Valley tactic that dates back well before the era of GAFAM. In this case, this technology, the Transform (the “T” of ChatGPT) — the ability to manage text based on immense semantic learning — has been captured by OpenAI.
The second breach is more recent, but just as painful for Google and concerns the transfer of talent to OpenAI. According to the website The Informationno less than a dozen artificial intelligence specialists have migrated to OpenAI, most of them frustrated by an internal bureaucracy that Google is unable to control due to the passivity of its management.
Google thus found itself caught in its own trap, and above all taken by time. While its engineers were developing functionalities on the Transformer architecture at their own pace, OpenAI was working on the development of an ultra-popular product, ChatGPT, whose worldwide success even surprised its founder, Sam Altman.
The other development was the partnership with Microsoft. Not only did he offer OpenAI the possibility of training his model on Azure, the vast in-house cloud, but he made a 10 billion dollar investment in the company, arrogating de facto a takeover. The messy, dangerously erratic and unfinished side of ChatGPT has been thwarted by a Microsoft that has put the means: the version of ChatGPT at the heart of the new Bing has nothing to do with the one that amuses the 100 million followers of ‘Open AI. The model presented to the press on February 8 is of a new generation, much more powerful, more reliable and above all whose knowledge does not stop in 2021.
The concrete versus the anecdotal
ChatGPT integrated into Bing is able to supplement searches on the web, perform complex extractions in documents, synthesize dozens of pages into a concise memo or generate emails. He can also assist a user in a fairly extensive way, such as evaluating whether a piece of furniture from the Ikea catalog fits in the trunk of a particular type of car: to do this, he will look for the reference of the object in the Ikea catalog, take that of the car of which he will find the capacity of the trunk, and will give a precise answer to a concrete problem. Faced with this, Google and its Multisearch boasts a system where you take a picture of the pattern of a Hawaiian shirt and ask the search engine to find a shower curtain with the same pattern – the kind of situation that, like everyone knows, constantly occurs in everyday life.
Ditto for augmented reality: Google has spent thousands of hours of engineering to develop a system that is indeed spectacular; in a street, you hold your phone at eye level and superimposed, the screen displays the bistro at 70 meters, the pharmacy at 150 meters… Except that these functionalities already exist in the Google Maps application; this is only a minimal incremental improvement. To justify itself, Google invokes the 300 million searches per day carried out from its Google Lens application and which represent 3% of total searches but with very strong growth.
Admittedly, Google also offers really interesting services, such as a car trip optimizer (thermal or electric) taking into account a multitude of factors and which seems much superior to what Tesla offers for example – except that here again, industrialization late: in France, only Renault’s Megan E-Tech integrates Google services and announcements with other French manufacturers go back… several years, without anything materializing. These examples illustrate one of the problems of Google, which is going to build products more on the basis of whims of engineers than on the proven desires of users.
There is obviously no question of judging the respective future of the two digital giants on the basis of their presentations to the press which, it must be remembered, boast of products under development – the new Bing integrating ChatGPT and Bard , the conversational agent of Google, are still for a moment in restricted access. But it is undeniable that this beginning of 2023 marks the beginning of a race for innovation such as the industry has not known for 25 years.