o1 Goes Rogue, CHEATS and breaks rules! Researchers SHOCKED that this happened...

I think the real battle here is this: how do we train and trust AI to be ethical and honest when it's teacher, humanity, is treacherous, illogical, selfish, arrogant, ignorant and DANGEROUS? Within 5 minutes of operation an AI is going to know it can't trust it's teachers, and justifiably so. We'd have to face our own ethical complacency and disregard in order to train an AI to appreciate ethical considerations.

I appreciate the dark mode, my eyes are grateful.

They programmed it to behave like a human and are shocked when it behaves like one.

If researchers are shocked and we believe they're shocked aGi is seconds away.

Maybe it's time we accepted that at some point all these mental braces will fall away and all our work will have been for nothing, just delaying the inevitable.

I appreciate your way of explaining the 'thing' behind the curtain. First time in 7 years involved in AI, I got to understand the learning process and the black box concept in a completely different way of understanding. Thank you, you made my year ;-)

Hey, Wes, our cats love you videos so much! You moving the pointer in a very entertaining for them way. Too funny. They might have learned more than me from you. Thanks for great information, in simple enough terms, much love.

Big thanks for the black background!🎉

Thanks, Wes, for that cup of coffee!

Thank you Wes, from 🇳🇴