Claude’s Emergency AI Warning is SCARY…

**Julian Goldie** (0:00)
I've given out a warning about AI, basically telling us that we need to stop, we need to stop using AI. The company that builds AI says when AI builds itself, there's some serious implications, which I'm gonna get into. So this is a brand new report from Anthropic directly. Some people, by the way, before we even get into this, are gonna say, listen, mate, they're great at PR, they're great at talking about stuff like mythos, and this is all a big PR campaign, which I can see where you're coming from. So we'll come on to that in a second, but you can actually see their timeline here in terms of how things are working. So they've broken it down, which is we had ChatGPT back in 20, now we've got coding agents, and you can see the progression here. It goes from literally coding agents, agents who were sometimes able to create code or create a nice website. We could do that last year, towards the end of last year, and today we have autonomous agents that can now run code themselves and delegate hours of work to other agents. And so the thing that we're looking at is that we are closing the loop, that things are getting pretty scary, things are warming up, and the rate at which AI models improve is accelerating. So the length of tasks they can run are running for hours and hours. We're looking at, for example, something where in 2027, which is less than a year away, of course, AI systems could be capable of tasks that take a person weeks. So you're seeing that progression between humans, where we were of ChatGPT and we were, yeah, ChatGPT is not that good. And now we're at the point where Claude is doing crazy powerful stuff and then we're kind of like tapping away on our keyboard and think, I'll give you an example. So earlier today, I was coding out the agent-operate system. I'm building out more stuff into it. We were building out Hermes Jarvis and this voice-activated system right here, right? Pretty amazing stuff, like it blew my mind.
And you know what I did? I just gave it one sentence and it builds everything. And I kind of just stood around feeling useless because I'm just waiting for Claude and my AI agents to build this out.
And in the meantime, I just stand there twiddling my thumbs because I don't need to have any input on this process, right? It just goes off and does something for me. And so we are seeing a serious progression. It does feel like it's moving fast. It feels like it's accelerating. It feels like it's getting scary and out of control. That's what we're seeing with Claude. I can understand the warning and where it's coming from because if you look at the rate that they're shipping and the crazy updates that they're creating, it is mind blowing, is scary, and it's only going to get worse, right? This is we're at the point now where it's not going to slow down, mate. This is getting crazy now. And I've seen it like I bear in mind, I can't code. But all of this is just built from me messaging an app and saying, build this out. That sounds like a cool idea. I bet you can't do that. And then it actually does it right. So you can see, for example, here, Hermes Jarvis, I can talk to it. And this is like a full activated voice agent that I've built just by going back and forth with an autonomous agent. And we've built that using Claude. And I can understand where they're coming from because where is this going? What is the point of us doing the work? What's the point of me even standing there when AI can create everything and we'll get to the point where it looks at me like I've got a peanut brain and it can do everything and automate everything without me.
And I think that's the way we're looking at it. And I actually saw something interesting here that was from an Anthropic employee, right? And you can see an example. So there's lots of cool quotes inside this. So for example, this Anthropic employee said, I started learning and leaning, sorry, leaning hard into Claudeifying about a year ago. That's been a crazy adventure. And it's now been five months since I last wrote any code myself. So a company that is a coding company, Anthropic, their team don't do any code.
And the other thing is, with these new updates, they've actually achieved 80% more code. And you can see the progression here. If you look at the code contributed per person by quarter, look at this trajectory here. And also look at how much Claude Mifos and Claude and everything else is bringing to the table, right? 8x more code with Claude Mifos preview. Now, the other thing to note here is again, like this could be hype because no one really has access to Claude Mifos. Apart from some crazy leak of a Chinese API the other day, which I've already done a video on, but the point here is like this is steepening in its acceleration. The other thing that I saw here as well, this is super interesting. So someone actually said they felt useless in so much words. I'm paraphrasing there, but you can see this other quote from an Anthropic employee, he says, on days where everything works well, I can't help but think nothing I do matters. Everything is automated and better and faster than I will ever be.