Scientists have discovered a quantum material that could represent the future of artificial intelligence – not because it retains vast amounts of data, but precisely because it doesn’t.
The human brain is often singled out as being the most complex and powerful computer that scientists know of, and one of the mechanisms that enables this complexity is our ability to forget things – a phenomenon that can be mimicked in a material called samarium nickelate.
“The brain has limited capacity, and it can only function efficiently because it is able to forget,” says one of the researchers, nanoscientist Subramanian Sankaranarayanan from the US Department of Energy’s Argonne National Laboratory (ANL).
“It’s hard to create a non-living material that shows a pattern resembling a kind of forgetfulness, but the specific material we were working with can actually mimic that kind of behaviour.”
The samarium nickelate the team investigated is an example of what’s called a quantum perovskite, an atomic crystal structure that starts to demonstrate changes in its electrical conductivity as its tampered with (by well-meaning scientists).
The tampering, in this case, comes down to simulations involving what’s called proton doping, with researchers exposing the material to hydrogen gas, which affects the composition of its crystal lattice.
“When scientists add or remove a proton (H+) from the perovskite (SmNiO3 (SNO)) lattice, the material’s atomic structure expands or contracts dramatically to accommodate it in a process called ‘lattice breathing,'” explains one of the team, Badri Narayanan from ANL.
These nano-scale gasps might not sound like the future of computing, but they could be what leads to it.
That’s because as this atomic respiration goes on, the material adapts, and its electronic properties start to change in response to the proton manipulations.
“Eventually, it becomes harder to make the perovskite ‘care’ if we are adding or removing a proton,” says one of the researchers, ANL physicist Hua Zhou.
“It’s like when you get very scared on a water slide the first time you go down, but each time after that you have less and less of a reaction.”
That kind of attitudinal change isn’t just a surge in confidence – it’s also a forgetting of fear – and in the case of the team’s manipulated material, which they call an “organismoid”, it suggests that technology could possess a memory that degrades and decays with time.
Just like how you can’t remember exactly what you had for breakfast six weeks ago – technological systems taking advantage of an electronic version of synaptic plasticity could mimic how our memory is sometimes unreliable, which funnily enough is actually one of its greatest strengths.
That’s because when we forget things, we make way for new, more important information in our memory banks.
This organic data disposal system is related to what scientists call habituation, a learning process in which organisms adapt to repeated stimuli as they become more accustomed to something, lessening their response to non-threats.
“Many animals, even organisms that don’t have a brain, possess this fundamental survival skill,” explains electrical engineer Kaushik Roy from Purdue University.
“And that’s why we call this organismic behaviour. If I see certain information on a regular basis, I get habituated, retaining memory of it. But if I haven’t seen such information over a long time, then it slowly starts decaying.”
Of course, just because researchers have observed this phenomenon in quantum simulations doesn’t mean we can instantly leverage the trick into smart, adapting devices straight away.
But the team thinks this habituated form of electrical conductivity could help to underpin new kinds of computing models that mimic our most important organ more perfectly than ever before – in all its imperfections.
“What we show in this paper is the extent of conduction and insulation can be very carefully tuned,” explains senior researcher and materials engineer Shriram Ramanathan from Purdue University.
“This could be really important because it’s one of the first examples of using quantum materials directly for solving a major problem in neural learning.”
The findings are reported in Nature Communications.