Lmao. Gen AI is causing outages at AWS and Amazon causing business income loss.

[capriciousday@mastodon.social]

@hyc @nixCraft too much investment has gone into the problem

[kitkat_blue@mastodon.social]

@nixCraft

ooh just *taste* that increased productivity!! :smacks lips:

[dianea@lgbtqia.space]

@nixCraft

The "Infinite Monkeys" Theorem in real life, AI

[rubinjoni@mastodon.social]

@nixCraft With all the talk about pushing stuff to prod on Fridays in the training data, I expect LLMs to push exclusively on Fridays...

[thejikz@infosec.exchange]

@nixCraft HAHAHAHAHAHAHAHAHAHAHAHAHAHAHA!!!

That is all.

[bazkie@beige.party]

@thejikz @nixCraft

[bazkie@beige.party]

@HankB @nixCraft I'm dreaming then.

LLMs can't ever be as good as senior coders at complex stuff, because they don't have as accurate an internal representation of the real world.

[htpcnz@mastodon.social]

@nixCraft saving time? That's if it was the goal, just like with pedo in chief and Iran war he started with axis of Epstein, assuming traditional goals of victory etc are why while everyone is laughing at them, they are laughing all the way to the bank going "look at these fools thinking saving time was our goal of adopting this nonsense"...

[hankb@fosstodon.org]

@bazkie @nixCraft

Today.

Tomorrow? Next month? Next year? Ten years from now?

[bazkie@beige.party]

@HankB @nixCraft well, I suppose first we're gonna have a massive bubble pop, which will stifle AI development for a while.

so my guess would be 15 to 30 years, but I wouldn't put any money on it

* to be clear; I'm not talking about LLMs here, they would need to design a more brain-like thing first

[pier@infosec.exchange]

@hankarus @nixCraft How will there be Senior Developers when they all quit because they have to review GenAI code all day?

[hankb@fosstodon.org]

@bazkie @nixCraft

> we're gonna have a massive bubble pop

That seems likely. One can hope for a softer correction but as frothy as AI has become, it could happen rapidly. And the results could be as you guess.

> they would need to design a more brain-like thing first

Perhaps. That *would* be monumental. However if LLMs work with words and we can use words to describe the world around us, that seems like movement in that direction.

[hankb@fosstodon.org]

@bazkie @nixCraft

We also need to decide what is "brain like" and that kind of introspection is difficult.

[patxi@mastodon.world]

@nixCraft
RIP senior devs, you will now work for the man and the machine.

[lumettaria@mastodon.social]

@nixCraft replacing coders with LLMs is starting to sound very expensive. Moreover, the junior-to-senior pipeline is broken now that the industry needs more senior devs for _code review_.

[bazkie@beige.party]

@HankB @nixCraft exactly, that's my thinking. they need a bunch of big breakthroughs, like transformers were for LLMs. but that takes time.

I believe LLMs are a dead end. using just words to describe the world is not multidimensional enough imho but that's just my intuition