Qualcomm is developing a new line of Zeroth processors designed to help devices think and sense just like human brains and bodies.
This isn’t just an attempt at mimicking the behavior of human intelligence, but rather, an attempt at replicating the underlying biology that drives behavior. Oh, and Qualcomm wants to stuff all that complex computation into a chip that fits in your phone.
Above, Qualcomm illustrates what it calls “biologically inspired learning,” which basically means replicating the process of how humans learn. In the video, the robots learn to go only to the white squares on a grid. It’s simple enough to teach a robot to do this with lines of code. The more complicated problem that Qualcomm manages to solve? Teaching the robots to go to the white squares by using positive reinforcement—building a ‘bot that can respond intelligently to stimuli in this way means that they can learn to do new things, even if they’ve never been explicitly told to. In order to interact and respond to the world the way a human does, Qualcomm has developed a complex models of neural networks that work by transmitting electrical pulses, much like human neurons.
Though the problem of replicating human brain function is quite the challenge on its own, Qualcomm wants to take it step further. Traditionally, researchers trying to replicate brains with complex computation have relied on massive supercomputers to do the number crunching. That’s all fine and well, but Qualcomm wants to build up an entirely new processor architecture that fits in power, and size, constrained spaces like a smartphone chip. Next to the central processing unit and graphics processing unit on your smartphone, Qualcomm’s Zeroth processors will have a neural processing unit to do brain like computation.
Now, this all sounds like a super fun science project, but the whole point of shrinking this type of function down to a mobile friendly size is to derive real tangible benefits for everyday people using the technology. According to the Qualcomm, the idea is that in the future, devices will see the world the way we do, so that they can anticipate our needs.