yes it's called "actually having democratic elections"

What does "democratic elections" even mean in this new world where traditional politicians don't understand these dynamics?

Then vote for politicians who do.

Few of those are ever running. Mostly we have just two brands of smooth-brains whose only policy aim is “preventing that other group of assholes from gaining any power, because they and their supporters are pure evil!”

Each side gets the smooth brains they crave because for ~50% of the population it's become a team sport / religion situation. A majority of people have not thought through 1/10th of the policy positions they automatically support/reject based on the team hat color.

There's a quote I will mangle and I forget the source of that's something like "If you agree with all the positions of your chosen political party, either you have thought through every option and came to the same conclusions on dozens of topics, or you haven't thought through anything".

Not a true democracy, then /s

also for ones fully into AI

too new, trinoo, ping -f, synflood, teardrop!

If only we had some way of signing messages

The technology isn't there yet (｡•́︿•̀｡)

Though in a case like this attackers would likely revoke (or publish) the private key.

Ah, perhaps we could put it on the blockchain! /s

delays between increasing capacity to produce and bringing production to market naturally lead to cycles of over and under production. https://thesystemsthinker.com/balancing-loops-with-delays/

where's a good irc chat these days?

It depends on the time of day, but #emacs, #nethack, #archlinux, #lobsters, #security, #openbsd usually have enough users for good convos. It depends on what you are into, really.

The develop-test-refine feedback loop for this kind of attack is so long (or expensive) that it seems likely to limit its real world use. Poison training data, wait months? a year? for the model to come out, see how well it worked, refine... or do you see a faster way to iterate?

Continual learning is the next major architectural milestone for the frontier labs. That’d reduce the iteration loop to days instead of years.

If your attacker assumes that all or most software will be generated from language models, the time penalty is worth paying.

if you don't know much math, it's easy to confuse the two

"what you see is all there is." it's generally much easier to verify something you've been made aware of than it is to know of it in the first place (and still verify it.)

The irony is that licensed interpreters / translators usually perform worse than AI.

Only the liability shifts from OpenAI to them.

Furthermore, where the alternative to a licensed professional was nothing, or a random untrained person or a weak professional, then it's harming the user on the pretext of protecting him.

(like in the other mentioned contexts).

did cursor do model training? I thought it used models built by other companies

Yes, https://cursor.com/docs/models/cursor-composer-2

It's fine tuned Kimi, they didn't train it from scratch.

Sure, but so what? It seems that model size and RL are the determining factors these days.