What comes after standalone LLMs? Causal inference and neurosymbolic architectures. "Probably correct" was always a liabilty in my field. So the key is to move from pattern matching ro rules and reasoning. #causality #neurosymbolic #explainableAI

And again. For critical systems, we'll need to move from LLMs toward neurosymbolic or other deterministic architectures. Probability is just not enough. #LLM ​#AI #Neurosymbolic #DeepLearning #MachineLearning #SystemDesign

It seems to me that the real goal of those big LLM players was never AGI, but a next-gen ad engine. Hard to imagine so many smart people in those companies genuinely believed transformers would take us to AGI. The AGI story was the hook all along. #AI #LLM #AGI #AIhype #BigTech

Funny how LinkedIn posts started sounding human again during the Cloudflare outage 🤔

For AI in medicine, the only way forward: AIs-in-the-loop, not human-in-the-loop. ​Meaning: A human is in charge and is supported by multiple, highly specialized AI algorithms. Not the other way around. ​#AIinMedicine #FutureOfHealth #HealthTech #MedTech #AI

Death Valley. Really does look like Mars. Just missing some @NASA and @SpaceX hardware. #DeathValley