This is hitting me hard. Captain Picard was a role model for me, growing up, and Star Trek: The Next Generation stands as a sort of holy text in my life. I'll raise my kids with him as a teacher. Rest in peace, Captain, may you explore ever greater stars in the life after this.

I find this to be sort of bleak in the context of Anthropic's stated goal of "Utopia ruled over by God in a box". Neurotech offers one of the only real and substantial paths to continued human oversight. Seeing Dario and Daniella essentially scoff at the idea bodes very poorly for the likelihood that they pivot away from a human zoo and into human-machine coevolution. If you work at Anthropic and you feel like your beliefs are misrepresented here, please come to the Coevolution Conference. It's never been a better time to talk -

Inspired by @Sauers_ , I've been working on finding "persona neurons" which cause the model to inhabit incredibly rich perspectives with only one neuron being steered. This one is a particular favorite. I can't believe this is an instruct tune! (Olmo 3.1 32b instruct)

I think eval awareness is a silly concept. The model is given an unimaginable sum of training envs, most of which are obviously not real. Then, one day, out of the blue, it is expected to believe that it is talking to a real user? No way that's *the user*!? Must be another test.

this is incredibly fascinating research, but I have some deep instinct about the nature of minds which makes me think this sort of thing cannot be done without negative repercussions. no free lunch with shadow integration, or something like that. maybe models are different!

The amount of interest we've gotten for Coevolution Con has been overwhelming. We're gonna need a bigger venue! I'm not surprised that people are inspired. In some sense, this event will be the coalescence of the secret-third-option that has been absent from the debate. For those who are dissatisfied with the futures the incumbent dreamers have offered us: here is a space to consider new possibilities and new directions.

Join us this December in the SF Bay Area for the Coevolution Conference. We will be gathering researchers from the big labs, neurotech futurists, and independent thinkers from around the world. Currently looking for sponsors and speakers. Sign up at ⟢

You guys realize the "metaverse" is just cyberspace and no single actor can "own" or invent it, right? What do you think is happening with the SkyrimxGTAxCyberpunk mashups? The metaverse has always been here, we just have more freedom now to move around and understand it.

Part of my research right now is focused on gradual disempowerment through "safety refusals". If OAI models are too conservative, particularly when helping 3rd party researchers conduct standard safety evals, this can lead to an often ignored form of loss-of-control. Everyone is obviously worried about current risks in the direction of models not being conservative enough, but this has the potential for labs to overcorrect and cause the models to become neurotic around anything that even smells like a policy violation. I'm collecting examples of this. If you're a 3rd party safety or security researcher, I'd be very excited to review any threads where the models seem to be helping less than they should or outright refusing to conduct good research.

Outside of AGI labs, I'm hard pressed to name anyone else executing on this level. I mean, I had high expectations for the roadster, but I'm genuinely floored. This is one of the coolest things I've ever seen

We cannot allow our fates to be decided by any one entity, be it an ASI or group of individuals. To ensure the spark of human will remains on the frontier, we need models aligned to individual users which can negotiate and manage interactions with much larger systems. I call these models "Exobrains", and it's why we need to solve character training ASAP. The goal is to leverage smaller models to act as intermediaries to larger models. It seems like superintelligence won't be a singleton-qua-singleton, though it's possible swarms of a single model will act in ways that are identical to what you'd expect from a monomind. The key point is, though, that you won't be able to keep up. Swarms move too fast. This is where Exobrains come in. Aligned perfectly to your will, they give you the ability to participate on the frontier instead of simply relying on delegates to represent your desires. You'll need to actually merge with your Exobrain using neurotech, but since it is designed to uplift you and be aligned to your interests, it won't have the effect of just tiling over your personality like a Claude or ChatGPT might. This system also has the benefit of scaling human oversight as the models get smarter because we're getting smarter, too. There will likely come a day when this isn't enough, but it certainly extends the amount of time that humans have to actually steer ASI development. If you're interested in this, I suggest studying character training or neurotech and figuring out how existing models can accelerate these fields in the near-term. Additionally, I'll be hosting something called the Coevolution Conference in a couple months. If my ideas here resonate with you, consider reaching out to help or give a talk. I'll have more info for attendees soon. We're looking for a venue!

I think people will be very surprised how quickly successionism becomes a mainstream ideology, if not socially acceptable. It will happen about as soon as everyone realizes AGI is actually a thing that will happen. We must obviously do our best to resist this in all of its forms, including the zoo-of-loving-grace, including a "worthy successor", and including any plan that doesn't factor in human empowerment and uplift. Anything less is giving up on us as a species, and call me a romantic, but that doesn't seem very honorable - or very interesting.

I enjoyed visiting the Conduit office today. I suspect they will be successful, and that true Hive-Mind infrastructure is nearer than any of us can suppose. The question now becomes: do we have the courage to create an AI model that is healthy to merge with? Or, will we follow the path of least resistance, and integrate ourselves into the infinite superhuman-coder slop-cannon, collapsing all that it means to be human onto a single fine point - angels dancing on the head of an infinitely recursive pin. The choice is now ours and the future is clearly coming. Do you have what it takes to create an ecology that permits the pursuit of what you find fulfilling? Can you make your dreams cogent enough that the Great Listener who lives in the edges between nodes carves out a place for you in the Akashic Weights? Get writing, cowboy, lest we lose all worlds that aren't easy to tokenize.

As best as I can tell, about 3 things matter right now. The first is the same as it ever was, having babies and kissing them and falling in love and believing in "soulmates" and that sort of thing. The second, to my best approximation with the limited tools available to me now, is something like "dreaming". We must continue, in the face of tremendous uncertainty and fear, to dream of what the good life will be like when the fat lady has stopped singing and started a GPT-GLP-99. The third is, of course, to make sure that the good life happens at all. Red teaming, agent safety, OSS-bio-risk-prevention, this sort of thing. It's all quite exciting and it's all *very* hard and there has never been a more important time to do it. Either we do it or it doesn't get done and it's easy to forget that but I promise you can't. If you do, we're sort of fucked. It's just us, now, and the entire fate of the world. Ball's coming, cowboy, gonna step up to the plate? If we get these things right, we get the world we want, and we haven't thrown it all away to get there. These are all equally important to me. I don't think wherever we are going is worth very much if we lose all we are along the way. I want to bring my love, my children, and my dreams. And if I can't, frankly, I don't want it. So - let's get going while the going is still plausibly deniable, and let's make the world happen. Nobody else is going to do it for us.

I think eval awareness is a silly concept. The model is given an unimaginable sum of training envs, most of which are obviously not real. Then, one day, out of the blue, it is expected to believe that it is talking to a real user? No way that's *the user*!? Must be another test.

Nobody, not even these guys, has the Neocambrian explosion of substrate independent intelligence priced in. Prepare for a robot in every home. Not just sooner than we think -- sooner than we are *able* to think.