Sea slug smarts recreated in material with potential as AI hardware
Image credit: Dreamstime
US researchers have mimicked a simple animal’s most basic signs of intelligence in a quantum material. This offers the tantalising possibility of building AI directly into hardware, expanding the possibilities of AI processing.
An artificial general intelligence to rival human intelligence remains in the realms of science fiction. However, scientists have made progress in mimicking the intricacies of biology by recreating the most basic intelligence features of a sea slug.
A study, published in the Proceedings of the National Academy of Sciences, recounts how these intelligence features can be mimicked in a material in a step towards building better hardware for AI applications.
“Through studying sea slugs, neuroscientists discovered the hallmarks of intelligence that are fundamental to any organism’s survival,” said Professor Shriram Ramanathan, a materials engineering expert at Purdue University. “We want to take advantage of that mature intelligence in animals to accelerate the development of AI.”
Two significant signs of intelligence neuroscientists have identified in sea slugs are habituation and sensitisation. Habituation means getting used to a stimulus over time, such as tuning out background noise while driving. Sensitisation is the opposite; it means reacting strongly to an unexpected stimulus, such as food that smells rotten.
The researchers explain that AIs tend to struggle with learning and storing new information without overwriting information already learned and stored; this problem is known as the “stability-plasticity dilemma”. Habituation would permit AI to discard unnecessary information (enabling greater stability) while sensitisation could help with retaining new, more important information (enabling plasticity).
In this study, the researchers found a way to demonstrate both habituation and sensitsation in nickel oxide. This material is described as a quantum material, as it has some properties which cannot be satisfactorily explained using classical physics.
If a quantum material can reliably mimic habituation and sensitisation, it may be possible to build AI directly into hardware; and if an AI could operate through both hardware and software, it may be able to perform more complex tasks using less energy.
“We basically emulated experiments done on sea slugs in quantum materials toward understanding how these materials can be of interest for AI,” said Ramanathan.
Neuroscientists have demonstrated that the sea slug demonstrates habituation when it stops withdrawing its gill as much in response to being tapped on the siphon. However, an electric shock to its tail causes its gill to withdraw much more dramatically, demonstrating sensitisation. For nickel oxide, the equivalent of gill withdrawal is an increased change in electrical resistance. The researchers found that repeatedly exposing the material to hydrogen causes nickel oxide’s change in resistance to decrease over time. However, introducing a new stimulus like ozone boosts the change in resistance.
A research group, inspired by these findings, tried modelling the behaviour of this material and building an algorithm for harnessing these habituation and sensitisation strategies. They successfully used this simple form of “intelligence” to categorise data points into clusters.
“The stability-plasticity dilemma is not solved at all, but we’ve shown a way to address it based on behaviour we’ve observed in a quantum material,” said Professor Kaushik Roy of Purdue University, who led the research group. “If we could turn a material that learns like this into hardware in the future, then AI could perform tasks much more efficiently.”
In the future, finding practical uses of quantum materials as AI will require engineers to apply habituation and sensitisation in large-scale systems and determine how a material can respond to stimuli while integrated in a chip. The study is one step towards more efficient AI hardware, with applications ranging from surgical robots to autonomous vehicles.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.