back
6 comments
Assuming by "AI" you mean LLMs in their current incarnation (starting with GPT), I tend to agree. On balance I see the harm outweighing the benefits for society.
what is AI? It is a loose word that changes meaning depending on the era.
I mean, there are AI applications that have improved humanity without threatening to enslave it, like computational photography, automated driver assistance for autos and biomedical applications.

LLM and stable diffusion though? Yep; agree.

Trad ML = good. LLMs have been a disaster for the human race.
Do you wish that nuclear weapons had never been invented?

My point is that like any technology, it's how you use it.

Absolutely yes.
> Do you wish that nuclear weapons had never been invented?

That's a very easy yes.

The wheel may have been a better example. It can be used for good or evil.

Nuclear weapons just destroy stuff at scale, by design.

Yes.
Well, a lot of us have a curious mind. Like, fission is a property of this universe. Gradient descent is a property of this universe. All you're saying is you'd rather not know about it.

I'm happy that nuclear weapons and AI have been invented, and I'm excited about the future.

Ok, but regardless of your feelings about AI, I don’t understand why you wouldn’t wish that nuclear weapons had never been invented. (Well, maybe it ended the combat between the US and Japan faster…, and maybe prevented the Cold War from becoming a hot war, but still, is that really worth the constant looming threat of nuclear Armageddon?)
If only having a curious mind would imply having a far-sighted and responsible one.
It normally does. That's why I can consider that nuclear weapons might have better uses in the future, presently unknown to us, and you can't.
Nuclear weapons could have a better use in the future? Pray tell, what exactly have you envisioned here?
My point is that you don't know what the future holds, but it's better to know more than less. My point is valid even if I can't provide examples.

However, if you ask me to, I can imagine using those weapons against meteors headeds for Earth, or possibly aliens. We don't know.

Phew, I never thought that "it's better to know more than less" would be controversial on HN.

The curious child takes apart an animal and learns surgery. The animal, however, is nonetheless killed.
I have a curious mind too but I don't go cutting up neighbourhood cats to see what they look like on the inside.
Well, you can learn what a cat looks like on the inside from a book. But someone did have to go around cutting up neighborhood cats, you're just benefiting from them. Which is the _whole reason_ why I maintain my position that inventing AI and nuclear weapons is a net positive for mankind.
If you're curious about that, are you curious about hypotheses like the Great Filter (Fermi paradox), and are you concerned that certain technologies could actually function as the filter?

I mean, what if the nuclear bomb actually did burn up the atmosphere? What if AI does turn into a runaway entity that eventually functions to serve its own purposes and comes to see humans the same way we see ants: as a sort of indifferent presence that's in the way of its goals?

There is a sort of people who read 1984 and blame the protagonist for being an idiot who called the fire upon himself, or still don't get what's wrong with ice9 and people behind it when turning the last page of Cat's Craddle.

And a sort of people who sympathize Winston and blame Felix Hoenikker, but still fail to see any parallels between "fiction" and life.

I don't know for certain if, when you say "a sort of people", you're referring to me, but... The sort of people you're describing sound like fascists, which is the opposite of me.
We're on the same side then, even if our opinions on the subject differ. Please take no offence.
I mean sure. I'm not saying "lets be wreckless". I'm saying "lets understand everything about everything, more rather than less".
I suppose I'd rather see us understand nuclear weapons without vaporizing civilians and causing an arms race. I'm not claiming to have a solution to preventing the arms race aspect of technology, but all the same: I'd rather these weapons weren't built.
It seems that most of the toxic AI stuff has really only been coming out of a handful of companies: OpenAI, Meta, Palantir, Anduril. I am aware this is a layman take.
If not them, it would just had been others doing the same thing.
And so? Then we'd be condemning those others instead. Just because others would step up to do terrible things if these companies didn't doesn't mean it's OK that these companies did.
ALL technology can be used for good or bad. It's the usage, not the invention.
It's real hard for me to conjure up "good" uses for, say, mustard gas or bioweapons or nuclear warheads.

"Technology is neutral" is a cop-out and should be seen as such. People should, at the very least, try to ask how people / society will make use of a technology and ask whether it should be developed/promoted.

We are all-too-often over-optimistic about how things will be used or position them being used in the best possible light rather than being realistic in how things will be used.

In a perfect world, people might only use AI responsibly and in ways that largely benefit mankind. We don't live in that perfect world, and it is/was predictable that AI would be used in the worst ways more than it's used in beneficial ones.

Why? Soviets tried to re-route rivers with nuclear blasts in their infinite scientifically-based wisdom and godlike hubris. How much illness their radioactive sandbox would cause among people was clearly too minuscle a problem for them to reflect on.

https://www.bbc.com/future/article/20250523-the-soviet-plan-...

> "Technology is neutral" is a cop-out and should be seen as such.

Uncle Ted, Neil Postman, Jacques Ellul were right all along.

While this is true, the horrific usage of a tool can vastly outweigh pitifully minimal benefits.

I'm not implying those adjectives apply to AI, but merely presenting a worse case scenario.

Dismissing the question of "does this benefit us?" with "it's just a tool" evokes Jurassic Park for me.

But not in equal quantity. Technology does not exist in a contextless void. Approaching technology with this sort of moral solipsism is horrifying, in my opinion.
I strongly disagree. Many technologies aren't neutral, with their virtue dependent on the use given them. Some technologies are close to neutral, but there are many that are either 1) designed for evil or 2) very vulnerable to misuse. For some of the latter, it'd be best if they'd never even been invented. An example of each kind of technology:

1) Rolling coal. It's hard for me to envision a publicly-available form of this technology that is virtuous. It sounds like it's mostly used to harass people and exert unmerited, abusive power over others. Hardly a morally-neutral technology.

2) Fentanyl. It surely has helpful uses, but maybe its misuse is so problematic that humanity might be significantly better off without the existence of this drug.

Maybe AI is morally neutral, but maybe it isn't.

Right. But a machine that helps plant seeds at scale could be used for bad by running someone over, but it's core purpose it to do something helpful. AI's core purpose isn't to do anything good right now. It's about how many jobs it can take, how many artists can it steal from and put out of work, and so on and so on. How many people die from computer mice each year? How many from guns? They're both technology and can be used for good or bad. To hand wave the difference away is dangerous and naive.
It only "take jobs" because it's useful. It's useful for making transcription at scale, text revision, marketing material, VFX, all those things. It also does other things that don't "take jobs", like computer voice control. It's just a tool, useful for everyone, and not harmful at all at its purpose. Comparing it to guns is just ridiculous.
But... the machine that plants seeds also takes away the livelihood to a bunch of folks. I mean, in my country, we were an agrarian society 100 years ago. I don't have the actual stats but it was close to 90% agrarian. Now, it's at about 5%. Sure, people found other jobs and that will likely be the case here. I will do the dishes while the AI will program.
I understand the industrial revolution happened. To say this revolution is the same and will produce the same benefits is already factually wrong. One revolution created a net positive of jobs. One has only taken jobs.
I would say we don't know that yet. Comparing the current state of LLMs to what they can lead to or what they might enable later on is like comparing early machine prototypes to what we have today.

I can also 100% tell you that the farming folk of 100 years ago also felt like the farming machines took away their jobs. They saw 0 positives. The ones that could (were young) went into industry, the others... well, at the same time we instituted pensions, which were of course paid for by the active population, so it kind of turned out ok in the end.

I do wonder, what will be the repercussions of this technology. It might turn into a dud or it might truly turn into a revolution.

> we don't know that yet

this is always the answer that the hopeful give: "previous revolutions of this kind, i.e., the Industrial Revolution, created a host of new professions we didn't even know existed, so this one will as well." Except that it was obvious quite early on what the professions created by those revolutions were. Factories were springing up immediately and employing those who had been working in the fields. The pace of change was much slower too; it took about 150 years in the US for the transformation from an agrarian society to an industrial one to happen. That provides time for society to adjust.

I have yet to see anyone demonstrate even an idea of new professions -- that may employ millions of people -- that are likely to emerge. So far, the "hope" is a pipe dream.

However, aren't there now a lot of job openings out there for LLM-whisperers and other kinds of AI domain experts? Surely these didn't exist in the same quantity 10 years ago.

(I'm just picking nits. I do agree that this "revolution" is not the same and will not necessarily produce the same benefits as the industrial revolution.)

We have very little idea what is "good or bad," especially over the long term.