Can ChatGPT eclipse Google or is it destined to become a great assistant for doing everyday digital tasks? After the discovery of the skill of the conversational robot doped with artificial intelligence from the company OpenAI, the time has already come for speculation about the future concrete applications of this platform phenomenon.
Available to the largest number since the end of November last year, ChatGPT has amazed technology specialists and the general public with the accuracy of his answers, his confusing retort and his ability to produce structured texts and contextual in all areas and in many languages. ChatGPT is the latest illustration of progress in so-called generative artificial intelligence. The algorithms in this language model are able to create new content by autonomously drawing from a huge open database – consisting of billions of web pages, Internet forum threads and literature – by structuring their responses and by adopting the language and style that desired by his interlocutor.
Ask him to write a philosophy essay for the matriculation exam, the introduction of a university thesis, an email in response to a supplier, a cooking recipe from the food in the fridge, to produce lines of code for an application, to invent a joke in the news or to compose a score… ChatGPT has the answeralthough it still has certain flaws, such as presenting false information as proven fact, not relying on the latest sources, or its ability to produce dangerous instructions or biased content.
Microsoft prioritizes integrating ChatGPT into its services
Behind ChatGPT we find the Californian company OpenAI, founded in 2015 and financed for one billion dollars by investors such as Elon Musk, Peter Thiel or Sam Altman to develop an artificial intelligence that benefits humanity. Launched as a nonprofit organization, OpenAI achieved limited profit status in 2019 to attract investors while continuing its ethical business.
One of OpenAi’s main investors is the giant Microsoft, which entered the company’s capital with a billion dollars and provided the servers for its Azure cloud-computing system needed to train ChatGPT and host its resource-intensive services. In turn, the inventor of Windows won the exclusivity of the chatbot’s commercial license and is now a priority to integrate it into its own services.
Convinced of the potential of ChatGPT, Microsoft would therefore work to integrate it into its Bing search engine to catch up with Google. According to sources cited by The Information, the idea would be to provide more human responses to requests from internet users. For example, if you are asked the question “When is the best time to go to Mexico?”, ChatGPT can provide a short summary, rich and well-argued, in addition to the series of links to the usual relevant websites.
Microsoft could integrate ChatGPT into its Bing search engine
However, this prospect, still very hypothetical, would be taken very seriously by Google. The undisputed leader in online research, with more than 90% market share, the Mountain View giant derives most of its revenue from this activity, which has changed little over the past decade. Google offers its results in the form of lists of links supplemented by commercial links that feed its powerful advertising ecosystem. The only deviation from this principle: the enriched results, in response to certain common queries, which are sometimes offered before the list of links, directly in the form of an information block, by subtracting from the database of the “knowledge graph”.
At present, the integration of ChatGPT with Bing does not seem to be able to challenge Google’s dominance in market and information research. But it could provide a competitive advantage to Microsoft’s search engine for all queries that require the mobilization of established knowledge and do not require additional navigation to an external link, such as definitions, recipes or historical landmarks.
According to the New York Times, CEO Sundar Pichai and the group’s leadership nevertheless mobilized and strengthened the teams in the face of a situation that resembles a “code red” internally. With the desire not to let a potential competitor take the land back. The giant has been working for several years on its own conversational robot, LaMBDA (Language Model for Dialog Applications), designed to be able to hold a conversation in human language, and which an engineer made headlines last year by claiming that this AI had become conscious , while publishing the transcript of a disturbing discussion about realism. But the program, still imperfect, has never been presented to the public.
ChatGPT could write your emails for you
Along with search, ChatGPT exploits can also benefit Microsoft’s Office 365 office suite. According to The information, the company’s engineers have already tested a version of ChatGPT in Word. “Microsoft engineers and researchers have been working for more than a year to create custom AI tools for writing emails and documents by applying OpenAI’s machine learning models to people’s private data. customers,” the news site said.
ChatGPT, which has already proven its ability to produce coherent and structured texts on the basis of short indications, could thus find itself in the end on Word, Outlook or PowerPoint to help users with their daily tasks : answering e-mails with only a few contextual indications, writing a cover letter to a very specific employer highlighting certain well-chosen arguments, finding and repairing broken links in an Excel workbook… The field of possibilities is wide among the tasks that can facilitate or even automated by the chatbot’s advanced algorithms while remaining under the supervision of a human user.
Microsoft would go a step further than Google here, which already makes it possible to auto-fill replies in Gmail with pre-written formulas inspired by the emails received. But ChatGPT should have improved significantly in reliability by then. Microsoft, for its part, will have to place the marker in the right place between the interest in advanced intelligent functions and the respect for the confidentiality of user exchanges. Currently, supporters of ChatGPT must accept that their use of the chatbot is monitored by OpenAI developers to gain access to the platform, because its algorithms must be constantly recalibrated and user responses can be used to correct its errors.
Meanwhile, millions of Internet users have already fallen in love with this extraordinary computer program, seduced by its ability to sort, summarize and clarify the requested information in record time. An enthusiasm that raises questions about possible deviations of its use while the robot is not yet connected to real-time Internet sources. Schools and universities have already banned its use by students, while experts worry about the opportunities it could offer cybercriminals to improve their online attacks or to trick people online by spreading fraud or false information. To limit these risks, OpenAI has indicated that it is working on a tool to automatically detect whether a text or part has been generated by the robot.