A Phone with a Mind of Its OwnBy
Despite appearances, the development of general intelligence in AI isn’t only happening on mainframe or quantum hardware with webs of networks in the cloud. Researchers are also looking at miniaturizing hardware specifically for portable devices like your smartphone.
You might think that your pocket computer already comes with advanced AI on board. Just hold down a button and Google or Siri will find all kinds of answers for you and even deliver them back in a number of languages. But the databases and engines for the virtual assistants are searched and processed somewhere away from your phone. Turn off the internet connection and your device will likely apologize for its inability to help at this time. As powerful as they are, our smartphones sometimes are just conduits to more intelligent computers elsewhere.
In order for our portables to catch up with the larger computers and networks, they require a number of mechanical miniaturizations, starting with the basic circuitry. One such important new electronic component called the memristor is being tailored for use in smaller devices. Its nano-sized elements and its curious ability to remember what it has done make it a perfect future fit for handheld computing.
SYNAPSES ON A CHIP
In the June 2020 Nature Nanotechnology journal, an MIT engineering team announced it had designed a “brain-on-a-chip” that was smaller than a piece of confetti. The chip contains tens of thousands of artificial neural synapses known as the aforementioned memristors, silicon-based components that can mimic the information-transmitting synapses in the human brain.
Remember how the transistor revolutionized early computers, shrinking the massive machines to console size? The memristor is now available for a next stage of miniaturization. The name is a portmanteau of memory and resistor. A switching kind of device, it differs from the conventional, linear resistor in its “dynamic relationship between current and voltage including a memory of past voltages or currents” (Wikipedia). Its discoverer, Leon Chua, called it a pinch switch because it could change conductance in a line dynamically. That “memory” it has of previous current flows and voltage is key for researchers who are building neuromorphic circuits today. A neuromorphic circuit is a large-scale system of integrated circuits that mimic the way neurobiological architectures move impulses in the human nervous system.
Chua didn’t actually invent the memristor. More accurately, he discovered the logical place in general theory of electronics for this kind of switch. That was 1971, and it wasn’t until 2008 that HP constructed one of these devices. Since then, a variety of memristors have been developed, all sharing the critical advantage of their nano-scale dimensions. The new MIT confetti-size chip has tens of thousands of memristors in a space that would be almost invisible in a portable computer. Two other important advantages of memristors include low-power consumption and a memory that isn’t volatile. The MIT researchers say their chip “was able to ‘remember’ stored images and reproduce them many times over, in versions that were crisper and cleaner compared with existing memristor designs made with other elements.” Their version was made from alloys of silver and copper, along with the silicon.
NEURAL NETWORK HARDWARE
The paper describes a second level of economy that would result from the similarity between a memristor and a synaptic connection in the brain. “Like a brain synapse, a memristor would also be able to ‘remember’ the value associated with a given current strength and produce the exact same signal the next time it receives a similar current. This could ensure that the answer to a complex equation, or the visual classification of an object, is reliable—a feat that normally involves multiple transistors and capacitors.”
Jeehwan Kim, associate professor of mechanical engineering at MIT, said in the announcement, “So far, artificial synapse networks exist as software. We’re trying to build real neural network hardware for portable artificial intelligence systems. Imagine connecting a neuromorphic device to a camera on your car, and having it recognize lights and objects and make a decision immediately, without having to connect to the internet. We hope to use energy-efficient memristors to do those tasks on-site, in real-time.”
The memristor is a significant step in the miniaturization of computerized AI. Today, specially designed memristors are available for computer memory chips, programmable circuits, and neuromorphic circuits for synapse projects.