AI has helped develop over 100 chips, should we be worried?
big picture: AI-powered chats and knowledge platforms such as ChatGPT, Google Bard, and Microsoft’s new AI-enabled Bing have received the lion’s share of attention lately, but this is not the only area where AI is making great strides.
Electronics design automation firm Synopsys recently reported that customers working with its Synopsis DSO.ai (Design Space Optimization AI) created 100 commercial feeds across a wide range of fields and nodes.
One client, STMicroelectronics, delivered the first ever commercial AI development stream in the cloud (in this case, Microsoft Azure). According to Philippe d’Audigier, director of SoC hardware development at STMicroelectronics, the system increased their power, performance, and area exploration (PPA) productivity by more than 3 times.
Junhyun Chun, head of SoC at SK Hynix, said In a recent project, the Synopsys design system reduced the cell area by 15 percent and reduced the die size by 5 percent. “Synopsys DSO.ai brings tremendous efficiencies to the development team by giving our engineers more time to create differentiated features for our next generation products,” Chun added.
In accordance with Synopsis, customers on average see more than 3x performance improvement, up to 25 percent reduction in overall power consumption, and significant die size reduction while using fewer resources. Having Synopsys DSO.ai on the team also automates many non-essential tasks, freeing up human employees to focus on other tasks.
Jean Boufarhat, corporate vice president of hardware and infrastructure engineering for Azure at Microsoft, said they are committed to democratizing advanced chip design, so hosting the Synopsys DSO.ai design system on Azure was a natural move.
An entertaining AI that creates funny images or (tries to) answer query-based questions is one thing, but having AI design hardware is another. What happens if something designed by an AI system malfunctions and injures or kills someone? Who are you pointing your finger at?
Even worse, what happens if the AI starts to work out hidden loopholes or actually becomes sentient? It’s great sci-fi stuff, but we’re fast approaching the point where it could be real, and it’s a little unsettling.