In this enlightening episode, we welcome Dr. Gottscho, a renowned expert in semiconductor technology and artificial intelligence, to unpack his groundbreaking study on the use of AI in streamlining semiconductor chip processes.
Dr. Gottscho delves into one of the major challenges in semiconductor chip production - the escalating cost and complexity involved in developing chemical plasma processes that form the transistors and memory storage cells. This process is currently manual, with expert engineers searching for a suitable combination of tool parameters that yield acceptable results on the silicon wafer.
The conversation takes an exciting turn as Dr. Gottscho discusses how AI, specifically Bayesian optimization algorithms, might help reduce the cost of developing complex semiconductor chip processes. He shares insights from his study that pitted human engineers against computer algorithms in a virtual process game designed for semiconductor fabrication.
The results? While human engineers excel in early-stage development, algorithms prove far more cost-efficient when nearing the precise tolerances of the target. Dr. Gottscho reveals that a synergistic strategy, pairing human expertise and AI algorithms, can potentially halve the cost-to-target compared to using only human designers.
Lastly, Dr. Gottscho highlights the cultural challenges inherent in blending human and machine efforts and discusses ways to navigate these when implementing AI in developing semiconductor processes.
Join us in this episode to explore the intersection of AI and semiconductor design. With Dr. Gottscho's expert perspective, we bridge the gap between humans and computers in the quest for more efficient chip production.
Kanarik, K.J., Osowiecki, W.T., Lu, Y.(. et al. Human–machine collaboration for improving semiconductor process development. Nature 616, 707–711 (2023). https://doi.org/10.1038/s41586-023-05773-7