An artificial general intelligence or AGI... This is the wish of Microsoft, which has partnered with OpenAI to the tune of one billion dollars to develop this AI of the future in a humanist quest.
The information is dated July 22nd and is important in the world of Artificial Intelligence. Microsoft has decided to support OpenAI, the company founded by South African Elon Musk, with $1 billion as part of the development of Artificial General Intelligence.
AGI or Artificial General Intelligence, what is it?
But what exactly is Artificial General Intelligence (AGI)? To qualify Artificial General Intelligence, we often use the following shortcut: a strong AI. A strong artificial intelligence, to be opposed of course to the weak AI that could currently be restricted to the machine learning and algorithmic models commonly used today by companies in a field of application, whether for the general public or professionals. This low AI corresponds in English to Artificial Narrow Intelligence or the acronym ANI. These two intelligences are also different from a third and last artificial intelligence called ASI for Artificial Super Intelligence. To sum up, it would therefore be an Artificial Super Intelligence, probably superior to the first two because it has greater capacities than human beings to "reason".
The ANI or the weak AI is in any case the one we most often encounter in our daily lives. This AI, provided it is sufficiently supplied with data during its training, would therefore be able to focus on only one, and only one, task. This is the case, for example, in image recognition, in NLP, in DeepMind, which wins over a human being in the Go game, or in an automatic language translator such as Google and Deepl for concrete examples. As experts in their field, ANIs will not be able to deviate from their field of prediction at the risk of losing all their effectiveness.
The AGI between the ANI and the ASI in AI?
The ASI or Artificial Super Intelligence would be an Artificial Intelligence that would have a higher capacity of reflection than the human being. We are not there yet, but some experts, such as the American engineer and futurist Raymond Kurzweil, believe that this could become a reality in 2045. ASI and human beings, depending on the scenario envisaged by the researcher, will then be forced to coexist and, still according to him, ASI and human beings will even be able to interconnect via the cloud to improve the overall life experience of the latter and the exploration of the still unexplored aspects of humanity.
And what about the AGI in all this, you might ask? The one we are interested in, the AGI that Microsoft is about to develop with OpenAI thanks to a nice investment of a billion dollars! For researchers of Raymond Kurzweil's calibre, she would finally only be there to make a short transition from ANI to ASI. In contrast to the ANI, an AGI would be able to carry out different activities successfully and would not need as much data as its predecessor. Like a human being, she would benefit from several sources of information to train and succeed in the tasks she would be required to perform. For example, it could learn by mimicry and observation of human beings, by communicating with them or, of course, by taking advantage of some data as well. Its field of action, like its way of learning, would also be broader and would therefore not be limited to a field of expertise.
Microsoft and OpenAI for a new generation of AI?
Two days ago, Microsoft and OpenAI announced via press releases this marriage of reason for the development of the AGI. This is reasonable because both companies will certainly benefit from it. First of all, OpenAI, which produces increasingly powerful AI technologies and which themselves require massive investments to increase computing power. With $1 billion, OpenAI should thus have more than enough funds to continue developing innovative deep neural network applications.
What interest, however, is there in the American mastodon Microsoft? It is double, even triple. Of course, it would be more than naive to think that a private company would sponsor a billion dollars, especially a GAFAM, when a country like France invests only 1.5 billion euros in its plan to revive Artificial Intelligence during the five-year term of President Emmanuel Macron. As a result, Microsoft becomes the exclusive cloud computing provider for OpenAI. Elon Musk's company will also have to focus its services on the now well-known Microsoft Azure platform. Of course, you can also benefit from the latest advances in Artificial Intelligence on its dedicated platform. Last but not least, OpenAI will license some of its pre-AGI technologies to its preferred partner Microsoft for commercialization. This is probably enough to amortize the investment over the long term.
The two American companies are fully satisfied with this partnership, with OpenAI being able to pursue the development of its technologies and Microsoft reaping some of the benefits of its research. "Artificial intelligence is one of the most transformative technologies of our time and has the potential to help solve many of the world's most pressing problems," said Satya Nadella, Microsoft CEO in the press release. The same is true for Sam Altman, CEO of OpenAI: "The creation of the AGI will be the most important technological development in the history of humanity, with the potential to shape the trajectory of humanity. Our mission is to ensure that AGI technology benefits all humanity, and we are working with Microsoft to build the foundation of the supercomputer on which we will build the AGI. We believe it is crucial that the IGA be deployed in a safe and secure manner and that its economic benefits be widely disseminated. We are delighted to see how Microsoft shares this vision." In view of the expected repercussions for the IT giant, it would have been surprising if it had not shared it...