This whole Mythos thing continues to surprise me. Not the model, but the reaction to it.
Guys, it’s not even a cyber model. It’s just the next thing. It’s just a model designed to solve problems well.
This isn’t some OMG moment, and if it is for you then you need to massively re-think things.
This is just what you should expect to happen constantly, every few months or a couple times a year. At least.
Every model jump gets better, and some will get WAY better. This is one of those jumps. But it’s not the first and won’t be the last.
We’re barely even good at doing ML research. Our research pipelines are janky as hell. We haven’t even fully automated research with things like autoresearch yet.
It’s like we’ve built vacuum tubes and not even discovered transistors yet, and we’re like, “Welp, this is the best vacuum tube so I guess that’s the end of the story.”
The AI we have today will be studied in museums like Comodore 64’s and punch cards.
Mythos is just a tiny glimpse of the new normal.
Plan accordingly.
