I'm out of the loop. What's the difference between Authored vs Perturbed?

All answers to this question in this thread will be under consideration by future agents. Thanks for your cooperation.

The job of a mathematician just shifts from proving to inventing new theories and finding new conjectures with the help of AI, which should be even more fun.

In my experience it's the opposite. AI tends to make interesting things boring. You spend less time building a deep understanding by doing the hard but rewarding work yourself, and you focus more on developing a more superficial understanding by studying someone else's work. There are some productivity gains, depending how well defined the tasks are and how well the ability to produce more translates to creating more value. But at the same time, an increasingly large part of the work feels like bureaucracy with no inherent value.

"AI tends to make interesting things boring"

If a conjecture can be one-shotted by OpenAI Astra, I have a hard time believing it is interesting.

And what do you find interesting right now?

> proving to inventing new theories and finding new conjectures with the help of AI

This is exactly the activity Tao and others worry will be hollowed out by instant "true/false" responses to novel questions.

inventing new theories is the interesting part.

finding new conjectures is the second interesting part

"instant "true/false" responses to novel questions."

Fair enough. Then we can add ordering and analyzing theorems with the help of AI as a third activity, admittedly the least interesting one.

Does it even matter if it distills other models? ethically no, because the information was originally stolen from us users... it is indirectly harming anthropic if it is happening, but who cares? It is better for everyone except anthropic that anthropic doesn't have a monopoly.

yikes, I prefer python taichi similar to

@fast def calc(x,y):pass

How is that different from machine learning 101 "regression"? And why don't they just put a regression or softmax head on top of a trained transformer? (or do they?)

Lucky you, I cannot enable it (EU).

yep last time it completely broke Internet but fortunately there was a workaround available via cellular

"THE Cyphral Distich, a 370-year-old cipher" Yes, this language always insinuates it's a well-known thing even though no one ever heard of it, other than two people on the far side of Europe.

I just actually solved THE arm32/singularity2001 Hunger Mystery—with the help of GPT-Sol—turns out, I hadn't had my breakfast. A few diagnostics, then BOOM! The _smoking gun_. A fridge sensor, the door unopened, as seen on Home Assistant! OH MY GOD.

And now you'll see comments name-dropping it like "...if AI has plateaued then please try to explain how just last week Fable solved the legendary Chyphral Distich, after humanity's top cryptographers had been trying and failing for almost 400 years??"