Next Generation AI Computing
summary
Claims that energy for AI communication could rival or even exceed world energy production lead to a push for hardware change, with CPUs/GPUs flagged as unsustainable and examples like stickers on stop signs causing autonomous cars to misclassify showing reliability failure. Results show some problems are noncomputable on digital hardware but become computable in the analog domain, prompting interest in analog computing and neuromorphic and biocomputing (IBM investing in neuromorphic chips). A Bavaria–Saxony funded Next Generation AI Computing project targets theory-driven hardware–software optimization and automatic EU AI Act verification for communication, medicine and robotics.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:51Computing in modern tech
- 2:34Energy and reliability challenges
- 3:58Computability problems
- 5:22Analog computing for AI
- 6:24Theory in sustainable AI
- 7:06Project goals
- 8:12Legal requirements for AI
- 8:56Bavaria and Saxony’s AI edge
- 9:58Vision for green AI
- 10:40Conclusion
related talks
01h 16mRethinking Computing with Neuro-Inspired Learning: Algorithms to Hardware
17mCan AI Become a Dystopian Threat to Humanity? – A Hardware...
21mAnalyst Notebook
19mComputing AI - Obstacles and Implications
43mKeeping Moore's law alive: Neuromorphic computing
22mCurrent State of AI
27mComputers for Super-Human AI
06mThe AI Compute Crisis & Web3 Opportunity