My @MLSysConf keynote is now online. https://
mlsys.org/virtual/2025/i
nvited-talk/2887
… The scaling of large language models has led to impressive gains in language understanding, but at a cost of insatiable memory and bandwidth requirements. I advocated a principled approach of designing optimization
RESEARCH
-
Optimizing Large Language Models: Memory and Bandwidth Solutions
By
–
-
XBai o4: New 32.8B Open Weights LLM from MetaStone AI
By
–
Here are my notes on XBai o4, the latest 32.8B open weights LLM to come out of an AI lab in China, this time from new-to-me MetaStone AI
-

AI Acceleration in Drug Discovery: Bottlenecks Beyond Algorithms
By
–
A good entry in the increasing number of articles that ask, given very good AI, what actually needs to happen for it to transform a complex field? Drug discovery is an area that is already seeing acceleration thanks to AI, but some bottlenecks are likely hard for AI to solve.
-
Structure and Observer Consciousness in AI Systems
By
–
If you stop paying attention to structure, the observer dissipates too
-
Automating Software Design with AI Mind Mapping and LLMs
By
–
Architecting With AI: Automating Software Design Using Mind Mapping And LLMs
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @demishassabis @Ronald_vanLoon @TamaraMcCleary @geoffreyhinton @goodfellow_ian @jeffdean @erikbryn http://
ow.ly/ooEz30sOj6Z -
Carbon-based brain meets silicon intelligence
By
–
Your inner brain is carbon-based, and your outer one is silicon.
-
Meta’s AI naming strategy reveals research priorities shift
By
–
How interesting that Meta's long-term AI research lab has "AI" in the name and its short-term one has "Superintelligence".
-
Quantum Mechanics and LLMs: Ugly Breakthroughs in Science
By
–
Ugliest breakthroughs in the history of science:
– Quantum mechanics
– Large language models -
GPT-5 Expected to Be a Major Breakthrough Model
By
–
Holy moly, GPT-5 will be a blockbuster of a model