Why Artificial Intelligence Researchers Should Be More Paranoid

There are always enough people who don’t care. I don’t want an AI to destroy humans, but frankly if it is an inevitability I’d rather it was an AI I designed.

WTF is this? These are and have been problems since forever. Yes, we now have to worry whether someone tries to murder you by replacing your roomba with a RoboWars KillerBot, contrived as it may be. Or even worse... Your voice and image may be FALSIFIED by unscrupulous actors and we are defenseless? There are still people alive from a time when any correspondence was written by hand, delivered by third parties, and the world kept turning.

The only way these contrivances would be harmful was if they were unknown to society at large.

Yeah, well, hold on to your hats because this is about to become messy.

And, I don't know why we continue to advocate for obscurity. It doesn't work, guys! But I agree that AI should become more like an engineering discipline, which would include training in best practices and safety.