IISc scientists develop brain-inspired computing platform

This technology could eventually bring sophisticated AI tasks right to our personal devices.

Researchers at the Indian Institute of Science (IISc) have unveiled a groundbreaking brain-inspired analog computing platform capable of managing an impressive 16,500 conductance states within a molecular film. This development, published on September 11 in the journal Nature, marks a significant advancement over traditional digital computers, which are restricted to just two data states.

The IISc team explains that this new technology could eventually bring sophisticated AI tasks—such as training large language models (LLMs)—right to our personal devices like laptops and smartphones, pushing us closer to making AI tool development more accessible.

Presently confined to resource-intensive data centres due to the need for high energy efficiency, these AI operations could soon see a revolutionary shift. As silicon electronics approach their limits, creating brain-like accelerators to work in tandem with silicon chips for faster, more efficient AI processing is becoming increasingly vital.

Sreetosh Goswami, an Assistant Professor at the Centre for Nano Science and Engineering (CeNSE) at IISc and the research team leader, shared, "Neuromorphic computing has had its fair share of unsolved challenges for over a decade. With this discovery, we have almost nailed the perfect system – a rare feat."

According to the IISc, such neuromorphic systems fundamentally perform operations akin to matrix multiplication—a simple concept taught in high school maths, yet one that consumes substantial energy in digital computations. The new platform by the IISc significantly reduces both the time and energy required, making these operations quicker and more efficient.

The platform's core, a molecular system designed by Visiting Professor Sreebrata Goswami at CeNSE, allows molecules and ions to create a vast array of memory states previously unreachable.