AI Slowdown Reverse Psychology

Trump is rage baiting AI so no one questions his low natural intelligence.

Well, one of Trump's professors said that Trump was the dumbest student he ever had. This shows that you don't need brains if you have a lot of money.

And people with less money listen to people like Trump because they have more money. AI cannot change that.

You speak like you never heard the term "regulatory capture"

Regulatory capture is when companies want the government to restrict every company in their industry, including them? Frontier models are the money maker, if they can't develop those it hurts their bottom line. And they've been talking about the dangers like forever, it's just that no one's listened until now.

The question mark tells all.

You're gonna have to be more specific about what your point is.

It is all a for profit scheme.

I'm saying you correctly identified regulatory capture.

I think it's one of those can of worms that once you open it, it's open, it doesn't fit back in the can when you want to close it again.

AI is a race to the finish line of fabulous wealth and power, self-regulated, government regulated, I don't think it matters, how do you monitor and restrict programming that happens behind closed doors, especially considering the regulators can't understand the programming.

I think we are asking them to please slowdown in your race to wealth and power, it won't hurt your business because, well, because we trust that your competitors will slow down too. You billionaires need to be on an honor system to make money slower. oh yeah, honor among billionaires, I don't see it happening.

Here's an idea, as always, follow the money. What if congress passes laws that make the AI company fiscally responsible for any resultant damage caused by their products. I don't think it's realistic to expect them to put the public's interest before their own financial interest, but if they will be held accountable when their product isn't controlled, then they will make sure they can control it. Considering the amount of damage thier product can do, making sure it doesn't happen is a matter of survival. Unfortunately, I don't think humanity's survival is motivating enough, but make it about the AI company's survival, and you get their attention.

It could cost lives! "Meh."

It could cost you all your wealth! "OK, ok, we must control this beast."

Regulatory capture only works if the regulations don't also harm the company calling for them. Most AI companies are running at a loss and their stock price is all in potential. If they can't develop frontier AI that does not help them and neither does a slowdown.

I don't think it matters, how do you monitor and restrict programming that happens behind closed doors, especially considering the regulators can't understand the programming.

You can monitor amount of compute and the type of training runs with zero-knowledge proofs. The people behind AI 2027 have written extensively about how this could work.

What if congress passes laws that make the AI company fiscally responsible for any resultant damage caused by their products.

Best case scenario, this bankrupts all frontier AI companies and causes a slowdown. But if they survive until an AI takeover, then saying "you have to pay a fine if you cause the end of the world" is kind of pointless, because it's still the end of the world.

Savant wrote:

Best case scenario, this bankrupts all frontier AI companies and causes a slowdown.

Isn't that what we are going for in this thread? You don't think AI can be eliminated do you?

Isn't that what we are going for in this thread?

Well I'd prefer a treaty with China that forces them to slow down too.

Savant wrote:

Isn't that what we are going for in this thread?

Well I'd prefer a treaty with China that forces them to slow down too.

Fat chance on that, I'm thinking they don't slow it down and get an AI takeover first, and their AI kills them and not us :)

Fat chance on that

The US is ahead, which means that at the current rate, AI will either (a) serve US interests, or more likely (b) go rogue. China doesn't like either of those, so it's in their interest to end the race.

their AI kills them and not us

I mean, this is less bad than most takeover scenarios but better for the AI not to kill anyone lol

My son told me there is a great documentary on Netflix that covers both sides of the issue very well, I'll go watch that and maybe will be informed when I get back.

It's called "How I Became an Apocaloptimist"

His short take on the two sides of the story, all the programmers working on it are warning us, all the CEOs getting rich off of it are saying, don't worry.

@Sidewalker

maybe will be informed when I get back.

I know that being informed about a subject is frowned on around here, but hey, I only said "maybe".

I'll go watch that and maybe will be informed when I get back.

Good on you for researching the topic. I remember it took me a while to learn how AI works and the doom arguments, and there's still parts I don't fully understand lol.

When AI no longer becomes profitable, it will be dumped and human greed will prevail.

I think that most folk here fail to address the bigger longer term picture.

The bubble or box of their meagre existence tends to be the limit of their consideration...Whereas Planet Earth has several billion years of development before it's final destruction...AI will be the key to the salvation of intellectual development and it's universal expansion...And clever people will be essential whilst the rest of us continue to fight over the crumbs of the Earth.

AI will be the key to the salvation of intellectual development and it's universal expansion

Sure, but it might be worth building the safety harness first to make sure it doesn't kill us.

With the US and China competing for AI leadership. It is hard to see any safety nets to slow the race.

There are verification mechanisms governments could use, but it would require the will to do so. If aliens came and threatened to blow up the solar system if we developed AGI, I think we would coordinate to stop it. So really it's a matter of political willingness.