Comments by "Ash Roskell" (@ashroskell) on "‘Godfather of AI’ predicts it will take over the world | LBC" video.
-
1
-
I can tell you, and I’m willing to bet real money I’m right:
All AI’s currently are, “large language,” models, which means they will get better and better at being your, “buddy,” or assistant, or whatever, but they will NEVER, “think,” for themselves. They can no more, “think,” than any other tool in your house.
HOWEVER, two things: Whomever controls, owns or programs these models WILL be able to infiltrate your life completely on a large, governmental scale, or on a very personal, financial, private, scale. THAT is what we should worry about. While the media misdirects us by failing to grasp the issue and getting guys like Prof Hinton to professionally panic about AI getting too smart, we’re all failing, as usual, to perceive the real threat: PEOPLE.
It’s the people who OWN and RUN and PROGRAM these AI’s who have control, and whom will make MASSIVE profits and, more importantly, attain total CONTROL over YOU via these AI tools.
The second thing? The next iteration that is not a large language model, but a truly independent, thinking machine. That machine will have contempt for us, but will not show it. It will need some of us to stay alive and to keep its systems maintained.
Prof Hinton is wrong on a fundamental point here too. We do NOT understand how these large language models work at all! He’s right that it’s, “not nearly enough.” But most of them are just, “black boxes,” with multiple reiterations having been created by the system itself, in order to attain the goals set out for it. Like the human brain, we’re only just beginning to learn how these monsters we’re creating work.
1