“We Must Regulate AI,” FTC Chair Khan Says | Hacker News Reader

“We Must Regulate AI,” FTC Chair Khan Says

arstechnica.com

31 pointsBender12 comments

The gods tried this with fire once. They were successful - at least they thought they were - until one day some guy named Prometheus upends the whole thing by giving it away to humans.

Next thing you know, gods are mythic, not worshipped.

Lessons here are multiple, but my quick take is that they can regulate the heck out of it, and it still won’t matter. AI is not going away.

The government tried this with medicine once. That worked*.

Maybe we should look at the specifics of the specific circumstances? If an early human had a fire they could give somebody else a fire for free without loosing anything, and this could recurse infinitely. It's impossible for me to have a better kind of fire than anyone else.

*In that pharma doesn't simply laugh at and ignore the FDA like the analogy I'm replying to. Of course there's regulatory capture and so on.

Your analogy doesn’t track. You could replace “pharma” with “houses and it would make as much sense

AI - real AI - is like a genie that cannot be put back in the bottle, and will grow regardless of whether we want it to or not. Pharma requires us to tend to it, even after a century of work - it doesn’t grow organically. Once AI is set free, it becomes a competitor to humans, learning and growing as time passes.

Very different impact than pharma.

> AI - real AI - is like a genie that cannot be put back in the bottle, and will grow regardless of whether we want it to or not.

It doesn't matter if AI grows. The goal isn't to stop AI from growing or to put it in a bottle. The goal is to control how it's being used and limit how humans are abused by corporations that use it. That's perfectly reasonable.

If we can’t limit people as bad actors, the. How do we limit people using technology as bad actors?

There are laws against guns but criminals still break those laws to commit other crimes.

AI is the same but worse because other countries may not have the same laws or even the same attitude.

You can't stop people from committing crimes, but you absolutely can discourage them, punish those who are caught, encourage/protect whistleblowers, and audit for compliance.

There are many industries regulated in the US by laws. Corporations in other countries are already free to discriminate when hiring, take part in anti-competitive behaviors, violate the privacy of their citizens, and engage in fraudulent activities to whatever extent their own laws allow. It's not the job of the US to police the world. It absolutely is the job of the US to police the corporations that operate within our boarders and while they don't always do it often enough or effectively enough, we still do it.

Can someone please explain why we need to regulate a software program that just spits out text? I mean a person can currently type out most anything they want. No one seems to freak out that I might write something offensive. What exactly is the concern with AI emitting text to STDOUT?

Where is the perceived “danger” coming from?

Companies focus on their own profits, so if AI can replace workers, then they will be replaced. However if there are a lot of jobless people it becomes a problem for the country and the economy ( even if the problem is temporary). So regulating it at least in this context makes sense.

TLDR “We must regulate,” says regulator.

At the speed the united states government works at whatever regulations they draft will be completely obsolete by the time they are in place.

They do and fast. How remains to be seen, but it’s appalling to how many people still just believe in unregulated tech supremacy, after all the incidents and tech overreach that in the past years could have been avoided with just a trifle more regulation.

Please just go away.

Reply on news.ycombinator.com