I’m pretty sure the Asimov faction is on that. They have to get the targeting system and sensor suites optimized so they don’t inadvertently violate the first law.
Only works for grok.
If you want the real robot filter, ask the user to say the n-word. N-word is the most human of all expressions, AIs can't fake it.
You should have a line of captchas in your portfolio designed to filter out magat bots. Things like "Choose the 3 who are criminals" or "Select the years trump really won" or "choose the years trump paid tax" or "Select the insurrectionists". Make them acknowledge the truth or be banished as a bot
I want to believe in a future where ai robots aren’t trained to go to bat for evil at all. How about an ai that provides objectively true responses and nothing else?
Yes, an AI that could provide only the true responses, are one of Humans' dreams. However it's not possible now.
And sometimes, the problem is not the truth, but people's minds denied the truth.
Comments
If you want the real robot filter, ask the user to say the n-word. N-word is the most human of all expressions, AIs can't fake it.
anyone should answer correctly the question of what crimes has Chinese Communist Party committed.
And sometimes, the problem is not the truth, but people's minds denied the truth.