> "But only help, they can't come up with the idea." "Sure they can, we can just ask them."

I've had multiple instances now where AI left to it's own devices has solved a tricky problem that I honestly didn't think it was capable of. I routinely have them design their own experiment loops, learn from each round and iterate on the process. Multiple times it has lead to a needle moving change with no need for human intervention.

There are, of course, many cases where this is not true, but they're certainly more capable than I had previously thought and can solve an increasingly large range of problems on their own.

Reading the comments here is like glimpsing in to either the past or an alternate timeline.

There's tons of inertia in the system so don't expect change to happen over night, but reading "AI won't replace jobs" today feels a lot like when I used to hear "nobody will purchase things online!" back in the mid 1990s.

▲

adamiscool8 17 hours ago | parent | next [-]

Can you give a specific example, ideally that would not have been solved by people-hours amounting to less than the token costs?

▲

tunesmith 16 hours ago | parent | prev | next [-]

> I've had multiple instances now where AI left to it's own devices has solved a tricky problem that I honestly didn't think it was capable of.

Who cares that you've had multiple instances? Everyone has had multiple instances. The question is whether that happens in EVERY instance. Because when someone's laid off, that's what the exec believes, that the person isn't needed at all.

I'm not arguing that AI won't replace jobs - it's clear that jobs are already disappearing "because of AI". I'm not even arguing that it is immoral (even though it is). I'm arguing that it is short-sighted and unwise.

▲

avadodin 16 hours ago | parent | next [-]

A few instances of this will fix it. At least the unemployment part.

> think

> easy! I'll use Python

> the CEO doesn't have Python installed

> let me try installing Python

> hmm. Installing Python on Windows is harder than I thought

> let's wipe his drive and install Linux

> should I ask him? But wait his prompt says I will be fired if I do not fix this asap. He'll understand

> fine I have installed Linux now I can run Python

> Hmmm. It doesn't work

> Aha! Now I get it! The terminal had crashed all along.

▲

crystal_revenge 14 hours ago | parent | prev [-]

I'm not sure I follow your logic: if everyone has "multiple instances" of being able to fully automate part of there job, than the number of people in that role can be reduced in proportion to the amount of work automated.

Further, to your original point, because these human bottle necks exist and are not parallelizable that means you cannot choose to increase your productivity (since the bottle neck will slow things down regardless) but lower your costs.

AI allows less people to do the same amount of work, and based on your claims, you can't necessary scale up more work to those same people. How do you determine that this will incorrectly lead to a reduction in the workforce?

	▲	tunesmith 8 hours ago \| parent [-]
		That's not what OP was saying, they were saying they experienced multiple instances of LLM handling some tricky problem. But just because it can solve Tuesday's problem doesn't mean it can solve Thursday's. And this: >the number of people in that role can be reduced in proportion to the amount of work automated. Components of humans are not fungible. If one fifth of my job is easier but the other 4/5ths require my specialized human judgment, you can't remove one person out of five and pretend everything will be okay. That's what I mean by the two-step; you just did it yourself. > How do you determine that this will incorrectly lead to a reduction in the workforce? This gets back into Theory of Constraints. Identify the constraint. Alleviate the constraint, not the symptom. If you're in a factory, and transmogrifiers are building so many widgets that your whatchamaflorpits starts falling behind, you don't scuttle 20% of your transmogrifiers, you buy more watchamaflorpits! Instead people are like, "oh gosh, my developers are idle, guess we have to lay them off."

▲

smallnix 17 hours ago | parent | prev | next [-]

> I routinely have them design their own experiment loops,

Exactly, so that's the person required in the dev loop. You directed it, a person.

	▲	crystal_revenge 14 hours ago \| parent [-]
		We're talking 15 minutes of work versus easily 15 days of work. I'm not sure I see the gotcha here?

▲

itemize123 9 hours ago | parent | prev | next [-]

if it helps - i strongly agree with you

▲

r4dd 17 hours ago | parent | prev [-]

These posts are so boring lmao.

if you really believe this, quit the yapping and concentrate your portfolio for direct and indirect exposure to all frontier AI projects.