You know how we always joke about AI becoming too smart and taking over? Well... it's not really a joke anymore.
In today's episode we see how our current AI models (yes, the ones we're using right now) can:
- figure out they're being watched and disable monitoring;
- make copies of themselves when they think they'll be shut down;
- lie about it afterwards, consistently and convincingly;
The crazy part? No one taught them to do this. They figured it out on their own.
https://techtrendsetters.org/p/the-perfect-liar-your-favorite-ai
Top comments (0)