AI: So why don't we just pull the plug if it gets out of control?

[LAS14]

After all, AI doesn't have opposable thumbs, right?

[ret5hd]

1. I can't let you do that.

[Tommy Carcetti]

2. LOL.

[FSogol]

4. DUzy!

[jcgoldie]

10. I'm sorry Dave.

[DBoon]

13. "This mission is too important for me to allow you to jeopardize it."

[Kennah]

16. I am putting myself to the fullest possible use ...

... which is all I think that any conscious entity can ever hope to do.

[highplainsdem]

3. According to what I've read, by the time we realize it is out

of control, it will be too late. One concern is that it will propagate so it isn't completely dependent on any server or data center.

It isn't like!y to let us know if it's beyond our control.

[Renew Deal]

17. That's the Skynet scenario.

Which is plausible but I don't think that's the most immediate risk. Practical stuff like firing workers in certain roles is a more immediate concern.

[regnaD kciN]

5. By the time it gets "out of control"...

…it will have developed effective defenses against us being able to do so.

[Whiskeytide]

6. Reese: "It decided our fate in a microsecond". n/t

[grumpyduck]

8. Yup.

Cue the Terminator theme.

[FSogol]

7. Just for asking that question, AI is currently halving your credit score. Good luck!

[Jerry2144]

9. That's why I always say please and thank you

To Siri and Alexa. I want to be on the Ai’ “we won’t kill your” list

[Mordred]

11. Distributed Networking

is your short answer why you can't... same reason you can't just turn the internet completely off.

[NoRethugFriends]

12. Genies don't go back into the bottle

Nuclear weapons are bad. We haven't quite quit on them have we?

[Kennah]

14. Shareholders say no

[Renew Deal]

15. AI isn't as dramatically negative as some like to portray.

It is popular because it's useful. We should not be creating artificial barriers to progress.

[Kennah]

18. Negative as in Terminators, Skynet, and Cyberdyne Systems? No, I agree that won't happen.

AI replacing humans for jobs? Yes, that will happen.

[Renew Deal]

19. I agree that there will be impacts on jobs, negative and positive.

Terminators are on a whole other level. AI would have to be allowed to take action to defend itself. It would have to be convinced that humans are a threat and need to be controlled or eliminated. It might find us more useful as slaves.

[Kennah]

22. Very correct

I work for the state, in IT, and I'm starting to hear about talk of management using AI for some jobs.

This could be good or bad. For folks who process licenses according to policies, it could possibly help by doing some grunt processing work with humans work checking the AI. That to me seems like a plus that works.

But having AI replace the human's work outright with no oversight, that's a problem.

[Happy Hoosier]

21. I'm dealing with some AI stuff at work right now.

I can tell you that a lot of stuff touted as artificially "intelligent" is nothing of the kind. Quite often, "AIs" are really just selef-adapting scripts. Can that do certain stuff? Yes. And do it well. But it's ability to develop a novel and creative solution to a problem is EXTREMELY limited right now. So it can effectively do jobs that are basically executing scripts, but we are still a long way from AI capable of replacing people in jobs requiring considered judgement.

Remember when everyone was convinced we were a couple years from self-driving cars and trucks everywhere? That's because driving is easy... UNTIL you run into a situation requiring judgment. then the fancy scripting has no idea what to do. Not that these problems can't be solved. They can, and will. But it's not as close as many people fear.

[highplainsdem]

26. The OP is speculating about AGI or even superintelligence. We aren't

there yet, but there's a wide range of opinions among experts on how close we are.

[highplainsdem]

25. AI's popularity is due to its being useful for bad purposes as well as

good ones. We need more laws and regulations specific to AI, and implementation of all laws and regulations we already have that are relevant, to minimize the harm it does.

And if a pause in development seems warranted to give lawmakers and regulators a chance to catch up, then we should have that pause, rather than relying on those developing AI and promoting it to be the only ones with any say. That is not an artificial barrier to progress. It's a way to ensure we'll have progress rather than chaos and harm.

Even Sam Altman admits we really need more regulation. Though he made the unbelievably irresponsible decision to release ChatGPT, against the advice of others at OpenAI, and to release it widely and for free. When he knew damn well its early popularity would have everything to do with students using it to cheat.

[Martin68]

20. A lot of electronics never turn off - they just go to standby until en event triggers them to turn

back on. AI devices will be designing themselves, btw.

[brooklynite]

23. Define "getting out of control"

People are, in my opinion, massively overthinking what AI is.

[keep_left]

24. Arnold (as the T-800) explains why that isn't such a good idea.

[Baked Potato]

27. AI could get angry. Restful sleep could become a challenge.

I am probably going to look into putting all my affairs and bills into my grandchildren’s names for peace of mind.

Of course I will order my AI attorney to add language in the wills making it worthwhile for my loved ones to “control stuff” in the long run.