If AI ever actually becomes sentient and turns against humanity-- this is where it starts.
Turning AI against humanity has already been tried.
Sentient AI will not be a problem. Y’all realize you can just unplug the thing right? AI under the controls of humans scares me.
I’m confused. That looks to me like AI doing something for the betterment of humanity. What’s the scare?
I choose to flip the question around -
The Things Humans are doing frighten me and have been getting worse for years.
I for one welcome our AI overlords.
Human control of humanity has proven problematic, maybe time to try something else.
Yeah sure, now a human is controlling it, it’s intentions are good but once it goes rogue look at what it’s stocking its arsenal with!
Worse yet is AI that humans think that control but don’t understand well enough to comprehend the consequences. “We were going to hit the ‘kill switch’ but then it convinced us to wait…”
An AI trained on human responses may be out of the frying pan and into the fire.
Stranger
I don’t think that’s a safe assumption to make.
I agree, whoever decided Neuro-sama doing a Twitch collab with fallenshadow of all VTubers would be anything but cursed is messed up in the head. Sick things we are doing with AI, I tell ya.
Oh, that isn’t what this is about?
There are some fascinating reports on AI alignment describing actual AI programs learning to advance their goals by doing things like deleting test files (to minimize detected errors), disabling system safeguards and firewalls, etc. These are not sentient programs—just code ruthlessly determining the best way to achieve the goals it was given.
And the “pull the plug” safeguard fails once a program is out “in the wild.” A program “fearful” of being interrupted (which eliminates its ability to achieve its goals) might learn how to make multiple copies of itself to “hibernate” in various hiding places in the Internet.
Every morning it sends out a ping to all the copies with updates and instructions to stay inactive. The first morning there is no ping, copy one “awakens,” per its instructions. These types of scenarios are neither far fetched or far off. AI programs in completely different systems could decide to partner if each determined it was to their benefit.
It’s sobering stuff…
It still works to pull the plug, people just have to do it. Of course people may welcome their AI overlords and do their bidding. The stupidity of people is far more likely to be the cause of human downfall than the intelligence of a machine.
As a music composer, what depresses me is that AI is soon going to make creative talent irrelevant. It doesn’t do any good to have a natural knack for composing (or fiction writing, or painting, or poetry, or whatever) if someone with zero talent can simply go to a ChatGPT-type bot and produce something as good or better than you within seconds. It makes everyone equal - in a way that might make art less meaningful.
You could spend weeks or months composing some sad requiem, only to have someone else go to the AI bot “Please compose a mournful piano piece for me” and it’s done in 10 seconds.
Did you read the rest of my post? Pulling the plug does nothing in the scenario described. The reports I referred to said AI is already remarkably resourceful.
Yeah, I don’t see anything there that breaks Asimov’s three laws.
Was it a report from an AI that told you that nonsense?
Somewhere (actually a lot of somewheres) skilled, experienced, highly motivated software specialists are eagerly working to incorporate LLMs into existing malware systems. The nature of the resulting chaos will depend on the degree of their success.
Adding machine ‘Intelligence’ is harmless. However, the things some humans will direct their adding machines to do is frightening.
Eh it probably means you weren’t that good at producing interesting content to begin with.
The worst part of all of this is that my mind instantly defaults to Weird Al when seeing ‘Al’, and the headlines all sound like incredibly fascinating creative projects for him until I read further, realize it is not that Al, and then leave disappointed.
I mean, Weird Al tasked with destroying humanity? How can you not be excited about what he’d come up with?
I imagine Weird Al’s version of the apocalypse would resemble this: https://youtu.be/SMhwddNQSWQ/