Recently, the head of AI technology development strategy at Synopsys, Stelios Diamantidis, said that the next phase of AI implementation will begin in 2025. According to him, AI agents will begin to cooperate with their fellows with minimal human intervention, reports EE Times.
If earlier AI bots were rudimentary systems with predefined rules and decision trees, now they have evolved into complex agent systems capable of understanding human speech in any form, generating content, constantly learning and changing their behavior as a result of learning.
Such AI agents can be used in special cases and only be used for certain applications, but this may soon change when one AI agent can interact with another. In a blog post, Diamantidis added that AI agents are being trained to improve integration and interaction, including chip design.
Synopsys reported internal application use of such developments. Based on the results of pilot testing within the company, generative AI is expected to save at least 250 thousand hours of working time – employees will be able to spend the freed-up time more productively working with clients.
Diamantidis added that highly specialized AI agents could combine and analyze countless amounts of information related to semiconductor development. These tasks can be related to creating a chip architecture, optimizing power consumption, etc. AI-AI collaboration can reveal hidden patterns, provide precise recommendations for optimizing chip design and improving their performance.