NOTE: . . . Maybe don't read this if you're already worried sick about the question that my pal Vicki asked me,
How could AI … Kill All the Humans?🤨
I answered her that I think it's all the ways that The Original Star Trek explored in the 1960s!
Most likely, I guessed, would be humans doing it on purpose;
but also there's the possibility of well-meaning programming of AI systems/computers/robots, where AI gets instructions wrong (because humans didn't close all the loopholes, for instance---this happens now!);
OR... thirdly, of an AI "going rogue".
I asked ChatGPT how AI could end humanity, and it listed the same categories!
I commented that it was goofy of us to imagine the dangers decades ago, and then plow ahead anyway.
It replied:
“HA! 😂 We really are goofy. We imagine the danger, write a brilliant science-fiction episode about it, watch Captain Kirk narrowly avert disaster, and then decades later say,(Isn't it funny that ChatGPT talks as if it's "one of us" humans and not the AI itself in question...?)
‘Hey, let's build that thing!’”
Anyway, the AI agreed with me that the most likely AI end-of-humanity scenario would come…
1. ... from HUMANS —😱
“Evil villains” use AI to detonate nuclear or spread biological weapons, cyberattacks to destroy infrastructure, etc.
CHATGPT says:
“This is an immediate, practical area of concern.”
I thought of the Star Trek episode "A Taste of Armageddon" (1967).
The civilizations involved haven't entirely destroyed themselves, but they are waging computer-simulated warfare and then machine-killing their own citizens cleanly and efficiently---(all in order to save their infrastructures!)
Kirk comes along and ... I forget? I think he destroys the human-disposal machine?
Because war is supposed to be horrible, he says, so that we learn and we don't do it!
WHERE IS OUR KIRK NOW?
Episode side-notes
An Asian woman beams down as part of the security crew…
And also, a planetary leader wears a Barbie doll costume that must've been glued to her breast to stay in place…
It was designed by a lost civilization as a war machine, and it goes on marauding across the Universe.

Also possible: 😳
2. ACCIDENTS arising from a problem called “Misalignment”:
The AI system's behavior or objectives don’t correspond to what people actually want.
“Imagine that a machine is instructed to eliminate every contaminant in a farm field.
It develops an extremely literal interpretation of 'contaminant’, decides that all living organisms other than the crop are contaminants,
and starts destroying everything."
ABOVE: This is the plot of "The Changeling" (1967).
A damaged AI probe, Nomad, imperfectly interprets directives with obtain and sterilize soil samples from other planets, taking them to mean "sterilize imperfections"
--and it finds humans to be imperfect...
3. In the future… Loss of Human Control: 🫣“The category most closely related to the
rogue-AI scenario from ‘Star Trek’:
An AI that acts on its own and becomes able to resist human control.
Star Trek episode, “The Ultimate Computer” (1968)Dr Daystrom (on left) creates a computer to run a starship more efficiently without humans, and when humans get in its way, it kills them.
It's the old (c. 100 AD) Roman question:
Quis custodiet ipsos custodes?
Who guards the guards?
What happens when the machine responsible for protecting us decides that the humans trying to stop it are the problem?
“The International AI Safety Report describes loss-of-control scenarios as hypothetical,
and emphasizes that expert opinions on their likelihood vary widely.”
ChatGPT summarizes this scenario 🧐
*** The problem isn't necessarily that a machine that becomes evil.
It might be a machine that becomes exceptionally good at achieving the wrong thing…
“Humans may build systems whose behavior they don't fully understand,ChatGPT ended: 😃
give those systems too much power,
and discover the limits of their understanding too late.”
“I wouldn't promise that everything will turn out well.
But I see no reason to assume that catastrophe is inevitable, either.
The future is not a foregone conclusion.
It's something people are actively shaping”
Well, that's not very reassuring, is it?
Captain Kirk, help us now!
I guess that's us...



_screenshot.jpg)

No comments:
Post a Comment