Problem Solving with Artificial Intelligence Requires Higher Compute Power
Artificial Intelligence Luminary: Pradeep Dubey, Intel Fellow and Director of the Parallel Computing Lab at Intel, discusses how artificial intelligence impacts a wide range of industries and is driving the need for massive amounts of compute power enabled by Intel.
Artificial Intelligence is simply about machines sensing, reasoning, acting, behaving like us human beings. It takes computing into areas where it has never been before.
Through computing we primarily solve problems which I call inside-out problems -- meaning someone gave you a very nice formulation of the problems. Generally nice equations. Newton gave you some equations. Einstein gave you some equations. Schroedinger gave you some equations. And then we just apply a lot of compute to solve those equations. But there are a whole bunch of other problems for which we are not lucky enough to have such nice formulations. What do we do with those problems? You start from outside-in. Meaning, you don’t know what the foundational equations are. All you know is, observations. Input-output. And 99 percent of the problems in the world are that kind of problem. These are areas like teaching, farming, health, education, politicking, decision-making in general.
Compute will actually improve the efficiency of these problems, these sectors of society which are operating way below the desired level of efficiency just to sustain our growth.
So what we’re talking now is really changing this perception that, yes, you need a lot of compute, not for solving tomorrow’s science problem, but you need a lot of compute to solve problems that we face every day.
And this is the reason why a compute company like Intel can actually be directly involved and directly cares about this. Because the need for compute has never been greater than what AI needs.
If we can actually deliver to the promise of AI, a whole new class of problems for which computing today does nothing, will suddenly become accessible to computing, and we’ll be able to make a difference.