It’s a good thoughtful piece. If I may suggest, the main problem and I mean the actual main problem are those people who underestimate the capabilities of AI. The ones who think it hallucinates all the time, unreliable and mostly useless.

The average person thinks this technology is overhyped (like Emily Bender) so how would one even begin to have a conversation with them on the risks?

▲

tao_oat 8 hours ago | parent | next [-]

This is my experience too. Many people in my offline, non-techy life seem to get almost upset when I argue that models are good at many things (and getting better fast).

And I kind of get it! If your primary exposure to AI is Google search summaries (from their cheapest models) and shitty AI graphic design, it makes sense to think it's vastly overhyped.

	▲	danofsteel32 8 hours ago \| parent \| next [-]
		I like how Tyler Cowen puts it: "This is the worst they will ever be". And some people are still stuck on what AI was capable of when GPT-3.5 was SOTA. They don't realize how much they've improved in ~4 years. Even if the rate of improvement slows down, AI will be tremendously more capable in another ~4 years.
	▲	simianwords 8 hours ago \| parent \| prev [-]
		Unfortunately it’s not just non technical people, it’s also technical people from my experience. They are usually too caught up in “how it always works” so they can’t think outside of it.

▲

justonepost2 8 hours ago | parent | prev | next [-]

I also think the "main problem" is people that annoy me on the internet

▲

ActionHank 8 hours ago | parent | prev | next [-]

It is overhyped.

It's powerful, and very useful in specific situations.

It's not the second coming of jesus christ.

	▲	mythrwy 7 hours ago \| parent [-]
		Agreed. It can code and we saw it eat(ing) our former occupation, but that was always a defined finite space with defined rules and to be frank, not as hard as us practitioners liked to believe. So I don't think we can extrapolate this progress out to every other human endeavor in the near term at least. More automation is coming though most likely.

▲

hattmall 8 hours ago | parent | prev | next [-]

Can you offer you argument you would provide to those people? I use the top SOTA models for software development and it is certainly an impressive and useful function. The quality of the code is however fairly poor, explicit architectural guardrails are frequently circumvented and the "human readable" outputs are very reminiscent of gish-gallop debate strategies.

Software development, as opposed to many many other fields also offers nearly unlimited training data and is one of the few areas with modifiable inputs and testable outputs.

▲

skrellm 8 hours ago | parent | prev | next [-]

> the actual main problem are those people who underestimate the capabilities of AI

Nope. The actual main problem are those people who totally ignore the real costs of AI.

Remind me, when was the last time when billions and trillions of dollars were spent on workers (education, healthcare, insurance, etc. all together)?

You try to compare the capabilities without comparing the costs. That won't do.

▲

ShutUpPedo 8 hours ago | parent | prev [-]

[flagged]

	▲	tytt 7 hours ago \| parent [-]
		He’s a strange creature I keep seeing his posts full of waffle and zero insight - it’s pollution for my brain