Microsoft CEO Pretty Sure He Can Keep AI from Escaping Human Control | Hacker News Reader

Microsoft CEO Pretty Sure He Can Keep AI from Escaping Human Control

futurism.com

18 pointskungfudoi18 comments

This interview was mostly a puff piece until this ominous quote:

> "The first set of categories in which we should use these powerful models are where humans, unambiguously, unquestionably, are in charge," Nadella said. "And so as long as we sort of start there, characterize these models, make these models more safe and over time much more explainable, then we can think about other forms of usage."

In other words, let's start with innocent chatbots until people get comfortable with them. Then we can start to do the menacing Big Tech stuff.

So first Microsoft will embrace this new technology and facilitate its employment, then they will start extending any established standards and provide new ways to use them, and it's anyone's guess what they will be doing next after that!

The next 50 years are going to suck. I'm already at the point where the Internet is basically becoming nothing but inaccurate information. I really try and find first party data, but it's hard.

Yeah, I've started to notice this in a lot of the internet, from articles to videos to podcasts. It feels like they are all selling me something I don't want, ideology, outrage, consumerism.

It's getting harder to find information or entertainment that's not embued with any of these negative attributes.

We're about as close to AI "escaping human control" as we are hosting the Super Bowl on Mars.

If there's one thing what we currently call "AI" has absolutely none of, it's actual intelligence. The rogue AI narrative helps companies pretend AI development is dramatically more advanced than it actually is.

Agree on the intelligence part. What it does have now, and at an increasing scale, is influence. So even if there is not, in the strict definition, intelligence behind the words, do not underestimate it's impact, both positive and negative.

At least Nadella is an engineer. I'd wager he has a better sense of this than ex-management-consultant Sundar Pichai.

Oh he's pretty sure, then that's fine I guess

Yes, they all said such things at the beginning. And then "Defense network computers. New... powerful... hooked into everything, trusted to run it all. They say it got smart, a new order of intelligence".

I think the only sad part is we'll have many Faux coupes' d'etat attributed by AI but in reality just executed by it's handlers.

AI will be the new scapegoat for everyone doing shady stuff, governments, corporations...

Or governments, corporations, etc., will be the scapegoat for everything done by AIs.

Touche

One can only hope. They need fire under their bums.

At least "pretty sure" is an honest framing, but nobody working on AI is going to say "well sure, it'll inevitably kill us all, it's pretty much a law of nature"

Well excuse me, but us humans are doing a great job at killing ourselves without the help of an AI thank you very much.

At this point it's the human control that I'm concerned about.

The great disconnect between AI safety guys and AI accelerationists is that the former produce blog posts and the latter produce AI.

Once again the doers beat the talkers. Shape rotators vs wordcels, if you'd like to use their lingo.

In author news, "Australian farmer "pretty sure" he can keep imported rabbit population under control."

Reply on news.ycombinator.com