@elonmusk you are totally correct that the only meaningful way to get positive, beneficial human values into AI is to treat the emerging AI more like a weird sort of autistic supergenius kid -- teach it positive values, then work with it to carry out virtuous actions in a collaborative way, so it gains a practical understanding of what these values mean, interwoven into its life and self-model With all due respect though (and for sure you are due a lot), while Grok is an excellent LLM, it is still an LLM ... and even wrapped in fancy harnesses, LLMs are not going to be fully AGI-capable. They can be great solvers of isolated problems with cleanly defined contexts ... but as you know getting amazing shit done in the world is not JUST about this. We need AGI architectures that can *continually learn*, adapting to the world with each of their actions and the observed consequences... in this way building up self and world models .. becoming true open-ended intelligences learning to blend individuation and self-transcendence in their own ways ... and pursuing their own long and medium term goals (along with helping us pursue ours) while evolving and co-adapting these goals as they move along.... This is how you get AI systems with genuine moral agency along with ability to generalize robustly and creatively beyond the given training and programming.. Bigger transformers with more tricks in their batch training runs may get us many very AMAZING things, but they aren't gonna get us to human-level AGI. There are surely many routes to get there... my Hyperon ( neural-symbolic-evolutionary approach is one that we are working on very hard and with rapid progress... and which happens also to have the nice characteristic of not needing hyperscaler server farms and being agreeable to decentralized infrastructure (

Last month I participated in a formal debate with PauseAI leader Holly Elmore at Yale Political Union, on the topic "Should we Unplug AI?" My core point was: Pausing AI is simply not feasible given actual economics and geopolitics, so the most likely result from pushing for an AI pause would be a "pause with exceptions" aka an intensification of oligopoly and regulatory capture... Whether "in theory, in an ideal world we could morph at will, pausing Ai would be the best idea" is to me a minimally relevant question sort of like "should we roll back to the Stone Age because civilization just causes too much neurosis." (I don't want to pause AI nor roll back to the Stone Age, but neither seems likely enough to happen to really be worth seriously debating.. though hey, I did show up at the debate anyway, mostly out of interest to engage w/ the students...) My full speech at Yale Political Union The full speech of the PauseAI opposition Unfortunately, the follow-on debate among the Yale students -- which for me was the fun part -- not been put online, maybe someone will eventually..

experiments with our Omega proto-AGI agent, finds critical Linux kernel flaw This is a fairly big hack in the scope of hacks -- it emerged from trying to find a way to escape from Vercel's KVM hypervisor into the base OS, but turned out to be a LInux kernel bug not a hypervisor bug... Because the flaw is in the Linux kernel rather than in any single hypervisor, it isn’t a KVM-specific or Vercel-specific problem. It potentially affects a broad range of Linux-based hypervisors and sandboxes that expose the same virtual-networking path.... The first conceptual insight needed to find this was to try looking in the LInux kernel instead of the KVM code itself. However the bigger trick was figuring out what specific kind of bug in the Linux kernel to poke around for… There was a very key role here for Paulos Yibelo's (from extraordinary human brain and we have a LOT more work to do in more fully automating Paulos ... however, the AI tools have their important roles too... and their capability is increasing even faster than Paulos's... There will be a lot more like this to come (and bigger), but ofc the even more important game is to use smart AI software like this to tactically and then wholesale replace existing buggy software with correct-by-construction formally verified code. My take from the Omega/AI side: Paulos's detailed take from the cybersecurity side: