Barbara Caputo * participates in the panel ‘The ways of artificial intelligence’ on 29 September at 12:20, in the Duomo.
Since October 29, 2021, Facebook transformed into Meta Platforms, the term metaverse, and the futurist virtual reality it evokes, it has become extremely popular and synonymous with parallel virtual reality: the three-dimensional virtual world inhabited by each of us through our own avatar, reached and experienced via wearable equipment and shared via the internet. This vision cannot be realized without virtual reality and without wearable sensors capable of making the experiences made by our avatars in the metaverse as vivid and rich as those experienced by each of us in our real life.
However, the role that Artificial Intelligence will play in all this is often underestimated: if it is true that in the near future the universe and metaverse will mix, that our avatars will interact with our physical selves – and with each other – in a continuous indistinguishable, then the times of the perception-action cycle in the metaverse must be the same as in the universe.
It cannot happen that if my avatar reaches out to touch a sweater in the metaverse, the feeling of softness of the wool comes to me with a delay of a few seconds, perhaps because the connection is slow. And then the metaverse will not be able to do without Ai models that anticipate, and if necessary simulate, the perceptive result of the action “touching the wool sweater”. That they make you feel on your fingers the feeling that you feel to touch a soft jersey, even if that signal has not yet arrived, perhaps due to connection problems or overloaded network. That starting from the gestures we start, they are able to predict what the most probable result of that gesture may be, and they “give us back” exactly that result, and no one else – even if maybe at the last moment we touched the silk blouse next to golf, or we stopped the gesture.
The AI ​​of the metaverse will therefore be closely linked to the concept of “embodiment”, even if virtual, and the development of new technology in this field will be closely linked with the development in the field of robotics and intelligent machines; with the hardware development of sensors and wearable processors, capable of making very small devices with low energy consumption intelligent, and at the same time with the ability to use gigantic amounts of data on increasingly powerful supercomputers capable of both generating the virtual meta-world , and to analyze the data generated by each of us as users of the metaverse. What will be the energy and environmental sustainability of all this? The Artificial Intelligence of the last ten years has developed by riding the growing availability of digital data, assuming that it has no limits of computing resources in its development. The planet’s climate crisis and the ongoing dramatic war are brutally confronting us with the fact that this model is no longer sustainable. The big challenge ahead of us is that of a sustainable Artificial Intelligence – will we be ready?
* Polytechnic of Turin