AI bug men: “It’s too dangerous !!! If anyone builds it will kill us all” Soviet Union: “Here is the complete patent for a thermonuclear weapon, by doing this we hope to foster innovation, collaboration, and contribute to other open source projects”

Anthropic has quietly built a biology lab in the Bay Area as it expands into physical experiments, Reuters reports. Its life sciences chief confirmed the wet lab. The company reportedly wants Claude to direct laboratory robots. Eric Kauderer-Abrams: "We're in the very early innings of using AI to automate the execution of lab work" Interestingly enough: "A spokesperson later clarified that Anthropic's lab is not for drug discovery specifically, declining to elaborate." This could mean that they are researching pharmaceuticals and, in addition, the fundamental principles of biology, such as aging itself. In any case, it is becoming increasingly clear, and Dario has repeatedly emphasized this, that they see biology as one of the key future fields for AI.

What do we know about Jacob Coxon, the Anthropic “whistleblower”? He became concerned about AGI doom in high school (in the town of Oxford btw - see my previous posts) after reading Nick Bostrom’s book Superintelligence. Attended Cambridge, during which time he went to an alignment conference at age 19. He then became a fellow at Newspeak (yes, really) House, which received EA funding and hosted Jeremy Corbyn’s Digital Democracy Manifesto launch. In 2022 he received a scholarship from a Moskovitz-funded org. Moskovitz is one of the leading EA donors, and an early Anthropic investor. The 80,000 Hours project (also funded indirectly by Moskovitz) explicitly encourages EAs to infiltrate frontier labs to influence them from the inside. None of this, in itself, proves that he is wrong about AI safety. What it does indicate is that he was already sold on the idea as a teenager, before he worked in the labs, before GPT-1 came out. I think good epistemic practice means we should therefore take him no more seriously than any other doomer. His experience in the labs didn’t convince him AI is dangerous. He already believed that even before the current AI paradigm became dominant, and probably chose his career based on this belief. If you’re convinced by Bostrom’s arguments yourself, that’s fine. But Coxon adds precisely zero additional weight to them.

OK ok, hear me out. Let's assume the the air gapped computer has no speaker. No way to emit sound. But the AI is able to run the CPU fast enough to kick on the fan. Which can can be picked up by any other computer within audio range (non air gapped computers will likely have microphones). That's at least a covert channel of one bit per ... uhhh ... 10min or so.

Jev is an AI model built for fast, low-cost decisions that went viral yesterday. It plays doom on its own because of low cost/latency decision. This will be the SOTA for World of Warcraft bots.

Self-replicating, multi-agent AI systems are an overlooked existential risk. I want to lay out why I believe this, via example. Each dot is a simple creature with a randomly evolved NN brain. There is no training; yet complex, unpredictable survival drives always emerge. 1/14

CANADA HAS BEEN INVITED TO BE AN ASSOCIATE MEMBER OF THE EU! "We want to bring the relationship with Canada to the highest level possible. We are opening the door for Canada to being the first associate member of the European Union"