Many science fiction works have been crafted on the idea that machines will become increasingly human. Machines that think, learn and make decisions in the same way humans do have been the object of speculative fear even as scientists and engineers work to create them.
While The Singularity is hardly looming on the horizon, there have been some fascinating developments in the world of artificial intelligence and machine learning that we wanted to delve into.
Researchers at Purdue University are building human brain-inspired hardware for artificial intelligence (AI) to help AI learn continuously over time. The goal of the project is to make AI more portable so that it can be used in isolated in environments such as in robots in space or for autonomous vehicles. By embedding AI directly into hardware rather than running it as software, these machines could operate more efficiently.
MIT engineers have designed a brain-inspired chip by putting tens of thousands of artificial brain synapses, or memristors, on just one chip that's smaller than one piece of confetti. Memristors (memory transistors) are silicon-based components that mimic the information-transmitting synapses of the human brain. This so-called "brain-on-a-chip" could one day be built into small, portable AI devices that could perform complex computational tasks currently only performed by supercomputers.
And researchers at Northwestern University and the University of Hong Kong have developed a device modeled after the human brain that simulates human learning. The device is able to learn by association via synaptic transistors that process and store information at the same time.
Here are more details on these three projects that aim to enable computers to mimic the human brain:
The Purdue Project
In a paper published in Science in February, Purdue researchers explained how computer chips could rewire themselves dynamically to take in new data as the brain does, enabling AI to continue learning over time.
To enable learning, the brain is continuously forming new connections between neurons. As such, to build a computer or machine inspired by the brain, the circuits on a computer chip also have to change. However, a circuit that a computer has been using for years is the same as the circuit that was built for the computer in the factory.
Consequently, researchers must be able to "continuously program, reprogram, and change the chip," according to Shriram Ramanathan, a professor in Purdue University’s School of Materials Engineering whose work involves discovering how materials could imitate the brain to improve computing.
Ramanathan and his team built new hardware that can be reprogrammed with electrical pulses on demand. The team's thinking is that because this device is adaptable, it will be able to take on all functions necessary to build a computer inspired by the human brain.
"Through simulations of the experimental data, the Purdue team’s collaborators at Santa Clara University and Portland State University showed that the internal physics of this device creates a dynamic structure for an artificial neural network that is able to more efficiently recognize electrocardiogram patterns and digits compared with static networks," according to Purdue. "This neural network uses 'reservoir computing,' which explains how different parts of a brain communicate and transfer information."
The team now aims to demonstrate this on large-scale test chips that could be leveraged to develop a brain-inspired computer, the researchers said.
MIT's 'Brain on a Chip'
MIT researchers are working toward the day when people "might be able to carry around artificial brains [that can work] without connecting to supercomputers, the Internet, or the cloud," according to a statement.
"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," the statement noted. "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."
In a paper published in "Nature Nanotechnology," the scientists explained how their brain-inspired chip could remember a gray-scale image of Captain America's shield — each pixel was matched to a corresponding memristor on the chip — and recreate the same crisp image of the shield numerous times.
In the paper, MIT's researchers highlighted the fact that its "brain-on-a-chip" could be used for performing complex tasks on mobile devices — tasks that currently only supercomputers can handle.
Brain-Like Device Simulates Human Learning
Taking a page from the book of Russian psychologist Ivan Pavlov, who trained dogs to associate the sound of a bell with food, researchers at Northwestern and the University of Hong Kong, have trained their computing device to associate light with pressure, according to a statement.. The research was published in the journal "Nature Communications."
This new device imitates the brain by using electrochemical "synaptic transistors" to process and store information at the same time, the statement noted. "These synapses enable the brain to work in a highly parallel, fault tolerant, and energy-efficient manner," the statement noted. The device's organic, plastic transistors work the way a biological synapse works.
"With its brain-like ability, the novel transistor and circuit could potentially overcome the limitations of traditional computing, including their energy-sapping hardware and limited ability to perform multiple tasks at the same time," according to the statement. "The brain-like device also has higher fault tolerance, continuing to operate smoothly even when some components fail."
In current computer systems, memory and logic are physically separated; however, bringing those functions together would save space and reduce the cost of energy. And the soft, plastic-like polymers of the new computing device would enable researchers to integrate it into smart robotics, wearable electronics, and even devices implanted in people.
While a robot uprising is still firmly in the realm of science fiction, these advancements are just some of the ways scientists are working to replicate the brain's biological machinery and that one day might lead to computers that function like the human brain.