How AI Will Reshape Computer Systems by 2035 (cra.org)

During a recent gathering hosted by CRA-Industry in San Francisco, prominent figures in AI, including Jeff Dean and Dave Patterson, discussed the future of computer systems by 2035. Sponsored by the Laude Institute, which announced new “AI moonshot” research awards, the event centered on how AI innovations will reshape not only computational architecture but also the way we design software and hardware. Participants expressed optimism that upcoming advancements could lead to a staggering increase in global AI inference—potentially 10,000 times more by 2031—driven by breakthroughs in algorithms, system optimization, and data center growth. Significantly, the discussions underscored the importance of both interpretability in AI models for human oversight and the need for advanced abstractions that may evolve beyond human comprehension. While trends indicate a profound shift in software development—where future developers might primarily manage AI agents rather than code directly—concerns regarding societal implications, including job displacement and tech backlash, were also raised. As the tech community grapples with these challenges, the insights from this Salon will inform the CRA's roadmap, emphasizing a collaborative approach to harnessing AI's benefits while addressing its risks in critical areas like education and healthcare.