'Youngest Korean Harvard professor' Ham Donhee says understanding the brain is key to future computing

By Park Hye-won, The Herald Business

Researchers have developed a semiconductor chip capable of measuring electrical signals in the human brain on a large scale, raising hopes for its use in next-generation neuromorphic chip research — an emerging field that seeks to replicate the brain's neural architecture in hardware to power AI functions such as deep learning.

The Chey Institute for Advanced Studies, chaired by Chey Tae-won, chairman of SK Group, hosted a special lecture by Ham Donhee, an endowed chair professor at Harvard University's School of Engineering and Applied Sciences, at the Korea Foundation for Advanced Studies building in Gangnam-gu, Seoul, on May 28. Ham graduated top of his class in physics from Seoul National University and joined Harvard's faculty in 2002 at age 28, making him the youngest Korean professor ever appointed at the university.

At the lecture, Ham introduced iMEA — short for Intracellular Microelectrode Array — a system his research team spent more than a decade developing. The system integrates roughly 4,000 electrodes onto a single chip to measure electrical signals from inside living neurons at scale.

In experiments on cultured rat neurons, the team simultaneously recorded intracellular signals from an average of 3,600 electrodes — about 90 percent of the total — and achieved a maximum of 3,900, or 97 percent. Using that data, the team successfully reconstructed a functional synaptic connectivity map comprising approximately 70,000 connections. The team also confirmed that the signals captured in the experiments reflected actual synaptic activity.

"This represents a significant expansion beyond the limits of existing technologies, both in the scale of synaptic connectivity and the precision of individual neuron data," Ham said. "Our goal is a new approach that can analyze large-scale neural networks while preserving cell-level information."

Ham said he expects the findings to serve as a foundational technology for designing next-generation neuromorphic chips — semiconductors that mimic the brain's neural structure in hardware to implement AI capabilities including deep learning.

"Today's computers keep memory and processing in separate units, but in the brain, memory and computation happen within a single network," he said. "Understanding how the brain works is directly tied to designing the computing paradigms of the future — it goes well beyond neuroscience."

A panel discussion followed, moderated by Shin Chang-hwan, a professor in the Department of Electrical and Electronic Engineering at Korea University. The two explored the industrial potential of neuromorphic semiconductors, covering what the brain's mechanisms of learning and memory could offer next-generation chip design and what new computing architectures might look like in the AI era.

"The brain's characteristics — integrating memory and computation, processing information at ultra-low power, and operating as a distributed network — can all serve as inspiration for future semiconductor design," Ham said.