OpenAI

“If we just publish the weights of a model on the internet, and then we realize, like there's actually a safety issue here, we can't take that back. It's done. It's like a one-way door. We also cannot put any usage restrictions on it after we open source it.”

“And the most positive long-term futures that I can imagine involve some degree of a merge. It doesn't have to be the crazy sci-fi, like, you know, plug something into our brains or upload ourselves or whatever. But at a minimum, I think it needs to involve som…”

“At the time, People thought we were crazy. Like, good researchers say you've totally discredited yourself by talking about AGI. Most other people were like, this is, you know, a hundred years away or more.”

“Eventually, now this will require more compute, but if we society at allocate to compute to that, you can say, like, could you go off and find a cure for cancer? And it can spend a lot of its compute cycles doing amazing scientific progress like that.”

“But I imagine that we get to a world where each of us has sort of like a, what feels like an AGI companion that we are talking to all day that is helping us be the best version of ourselves, learn, be efficient, be happier, and we'll all experience that way.”

“The other company I'm involved with is this nuclear fusion company called Helion. And there are a lot of similarities between the two companies in terms of like a very hard Scientific and engineering problem. But the biggest difference is like what to do, what…”

“And I think if we're able to accomplish that, it will be perhaps the most important invention in human history, and it will be the culmination of this phenomenal amount of effort all the way down the stack and like the collective knowledge, accomplishment, mor…”

“And we have something called the merge and assist clause, which says, similar to that, if another effort's ahead, and we want to avoid a race condition we can just shut down and merge with some other effort.”

“And I think the same thing will happen for AI. There will be a layer of intelligence provided by us and others. It's just everywhere. And as the systems continually get smarter, everything you use will get smarter and smarter as well.”

“Well, it turned out to be a mistake. I mentioned that Dota was a great environment. It turns out that robotics are a very hard environment. Robotics is hard, and it's not hard because the machine learning is hard. It's hard because the simulator is bad, the ro…”

“Right now, you know, we deploy our technology slowly and cautiously, like we're willing to piss users off to go slowly, but as the systems get much more powerful, the challenges become more and more unprecedented.”

“The way we pick projects to work on is, like, not as exciting as everybody would hope. You know, I think there's this belief that you have a bunch of people, like, sitting in a room picking this, like, brilliant secret strategy. We just sort of, like, run our …”

“We did also just see if the government wanted to fund us. They definitely did not.”

“We also have something in our documents which says if we need to for safety, the board can just totally wipe out everyone's equity value to nothing.”

“The first thing we did is we sat in an org-wide NDA, and I said, hey, how many folks in the organization can I get, we give access to? And eventually we got about 40 or 50 folks in our organization to get access to GPT-IV.”

“Especially in those early days, it was Still pretty bad at math.”

“Remember that the stated mission statement of OpenAI, Anthropic, and DeepMind from the very beginning, those three companies, Have been to create artificial general intelligence.”

“Our architecture allows for multiple models. Things are made by companies like OpenAI. We have a good relationship with the team there and smaller models of things that are proprietary to us, and these will change and evolve over time, and we are able to Add t…”

“Another company, OpenAI, latched onto the idea that this is, in fact, extremely scalable, more scalable than anybody had appreciated. And they just basically scaled it up and scaled it up and scaled it up.”