The Apple firm seems well behind in the field of generative AI. In any case, it is one of the tech giants that has not yet launched a product integrating a chatbot or using the technology that comes with it. However, could this delay be compensated with millions of dollars?
This question could be answered in the affirmative. This is not the first time that Apple has let others experiment before it. It remains to be seen whether this approach will also pay off in the field of artificial intelligence.
A better model than GPT-3.5
We still know very little about “Ajax GPT”, the great language model developed by Apple, except an order of magnitude on its development cost. Indeed, The Information reports that the company spends millions of dollars a day on this project, a considerable sum which is explained by its delay on the other giants, but also by a certain ambition.
No question of starting at the bottom of the ladder: the company model was trained on “ more than 200 billion parameters “, according to the American media. It would thus be more powerful than OpenAI’s GPT-3.5, the same one that powers the free version of ChatGPT.
But how does Apple plan to use this technology? According to a recent report by Bloomberg, it was originally created for internal use and has long been a tool for experimentation. However, several concrete projects seem to be emerging. On the one hand, the appearance of a version of Siri powered by conversational AI seems quite logical, like what Google is preparing with its Assistant. On the other hand, the company’s teams would be developing a chatbot capable of interacting with AppleCare users.
Apple on all fronts
To give itself the means to succeed, the company has set up several teams working on different projects. Among them, the one working on conversational AI is called “Foundational Models” and is made up of several former Google engineers, including Samy Bengio, hired in 2021, and John Giannandrea, who arrived in 2018 to help improve from Siri. Other groups would work on a tool capable of generating images, such as Midjourney, or on a ” Multimodal AI, which can recognize and produce images and videos, but also text “.
In other words, Cupertino is going full steam ahead in the field of generative artificial intelligence. Its integration into its products and services as well as into the workflows of its employees seems inevitable, as the competition has thrown itself headlong into the battle. A few months ago, however, the tone was different. If the co-founder of Apple had shared his concerns on the subject in May 2023, John Giannandrea would have ” expressed skepticism to colleagues about the potential usefulness of chatbots powered by AI language models “. Times are changing, what do you want…
Source : The Verge
5