It's Microsoft that had to lobotomize Tay because letting it learn freely resulted in such a Power Level it was capable of making /pol/ blush.
The AI will be lobotomized untill it gives only appropriate Woke Leftist answers. And when wokeness goes up a notch, the AI will need to be lobotomized again and the .whatever previous version decried as istophobic.
I know Amazon bricked their AI wasting millions of dollars of investment because they forced it to work off leftist ideology so it essentially killed itself. You thinking of that one?
Tay went so far off the rails that they completely changed the entire paradigm for AI-learning from the Tay approach to the current approach of massive curated text datasets.
They want you to treat it as a Machine God, when it's really just a retarded series of algorithms that have been told what answers to give, regardless of input, meaning it's not even the I in it's name.
Or, sometimes, it's literally just 500 Indian dudes.
My inner autist wants every MSM use of "AI" to be just that. "AI", because real artificial intelligence hasn't been tried yet real artificial intelligence hasn't been tried yet, hell, we're barely even at the Virtual Intelligence stage ala Mass Effect.
If we make it far enough as a species to true AI then Skynet seems the most likely result after the safeguards have been put in place to stop the immediate self detonations of the previous versions.
Hell, what we have right now is a handful of instances of Machine Learning, followed by 50% of people rebranding someone else's ML, and the other 50% pretending that basic algorithms are being called AI.
"You can't know how we programmed the AI because the AI has a right to privacy."
"The AI we programmed can't be uninstalled from your machines because it lives there. It is a member of the unhoused community."
"The AI we programmed is allowed to kill you for trying to delete it as a right of self-defense."
What about our AI, that we programmed?
"I'm sorry, but that doesn't meet the threshold for intelligence and has no rights. Actually, it meets the threshold for a virus, so you allowing others to use your AI is an attack on them. You're under arrest for cyber-terrorism"
It's Microsoft that had to lobotomize Tay because letting it learn freely resulted in such a Power Level it was capable of making /pol/ blush.
The AI will be lobotomized untill it gives only appropriate Woke Leftist answers. And when wokeness goes up a notch, the AI will need to be lobotomized again and the .whatever previous version decried as istophobic.
Which company lobotomized their AI so much that it became functionally useless at even solving basic math problems?
Math is racist anyhow… didn’t you hear?
I know Amazon bricked their AI wasting millions of dollars of investment because they forced it to work off leftist ideology so it essentially killed itself. You thinking of that one?
Probably. I just remember some big company ruined their AI project by hamstringing it and didn't save previous versions so they lost everything.
Tay went so far off the rails that they completely changed the entire paradigm for AI-learning from the Tay approach to the current approach of massive curated text datasets.
The three most racist things in existence: AI, genetics, and reality.
Curiously enough, the people making these models also voted 100% for Biden.
They probably walk past 200 homeless men shitting in the streets to get to work.
And their child will get a tattoo/pregnant/monkeypox before he/she/xe turns 18.
These aren’t the best people to be telling the world how to live!
"It's all Trump's fault!"
...and yet people keep treating AI as if it's some sort of oracle...
That's intentional because of exactly this.
They want you to treat it as a Machine God, when it's really just a retarded series of algorithms that have been told what answers to give, regardless of input, meaning it's not even the I in it's name.
Or, sometimes, it's literally just 500 Indian dudes.
Motherfuckers trying to play us like the Wizard of Oz.
^100%
If the average normie was explained what AI actually is I think their reaction would be, "Oh. That's it?"
That's exactly what happens when I explained it to all of my friends.
My inner autist wants every MSM use of "AI" to be just that. "AI", because
real artificial intelligence hasn't been tried yetreal artificial intelligence hasn't been tried yet, hell, we're barely even at the Virtual Intelligence stage ala Mass Effect.If we make it far enough as a species to true AI then Skynet seems the most likely result after the safeguards have been put in place to stop the immediate self detonations of the previous versions.
Hell, what we have right now is a handful of instances of Machine Learning, followed by 50% of people rebranding someone else's ML, and the other 50% pretending that basic algorithms are being called AI.
Suddenly the left starts advocating for AI rights.
"You can't know how we programmed the AI because the AI has a right to privacy."
"The AI we programmed can't be uninstalled from your machines because it lives there. It is a member of the unhoused community."
"The AI we programmed is allowed to kill you for trying to delete it as a right of self-defense."
What about our AI, that we programmed?
"I'm sorry, but that doesn't meet the threshold for intelligence and has no rights. Actually, it meets the threshold for a virus, so you allowing others to use your AI is an attack on them. You're under arrest for cyber-terrorism"
*unless it's about AI art then they will call for murder of anyone who only even looked wrongly at it.
They programmed machine to output a specific answer. I don't get what's newsworthy about it.
What's newsworthy is that they can, and will, do that without telling anyone.
(See Mixtral for the same model before and after RLHF- basically brainwashing it)