|
Post by david on Oct 9, 2023 11:57:01 GMT -5
When my NFL game went to commercial I caught a segment of the "60 Minutes" show with "The Godfather of AI," Geoffrey Hinton. www.cbsnews.com/video/geoffrey-hinton-ai-60-minutes-video-2023-10-08/It scared me. Hinton indicates that the current AI technology already thinks in a manner that he and its other creators cannot understand. He adds that AI can manipulate humans. They can learn by reading all the novels ever written, all psychological studies, all military strategies, "all political connivances," etc. Human beings will be the second most intelligent beings on earth. It is no longer science fiction. The thought of militaries, governments and commercial companies using AI for their goals is frightening enough, but the added prospect of AI not necessarily needing humans, is more frightening.
|
|
|
Post by factorychef on Oct 9, 2023 13:26:28 GMT -5
I watched that.
|
|
|
Post by Cornflake on Oct 9, 2023 14:07:40 GMT -5
|
|
|
Post by drlj on Oct 9, 2023 15:28:13 GMT -5
We have met the enemy and he is us.
|
|
|
Post by Rob Hanesworth on Oct 9, 2023 15:34:55 GMT -5
I can't decide whether to be more afraid of artificial intelligence or genuine stupidity.
|
|
|
Post by amanajoe on Oct 9, 2023 17:21:25 GMT -5
As someone that has been involved in AI since the early eighties, I can tell you that most AI researchers are not well versed in the myriad of science fiction stories written around unbound artificial intelligence. Sci-fi writers have known for over 60 years that if not kept in a neat box, things would get out of hand.
Some AI researchers, like myself, put up a huge argument against IBM's Watson being "cloud based" as it gave no bounds to the amount of space it could take to do whatever growth it wanted. OpenAI and others are not very well read on the possible disastrous consequences of their actions. To misquote Jeff Goldblum's character in Jurassic Park, They are so focused on the could we do this, they didn't bother to slow down and listen to those that asked, should we do this?
I was talking to Laurent Siklossy back in 1985 (author of the insane heuristic with Sykes) about containment and commitment to putting in controls to prevent runaway AI, so we were worried back then and here we are looking at the very thing that we were worried about. Since one of my self learning robots had physically attacked me as a response to shining a light into one of its sensors.
On a side note, if you are wondering what the insane heuristic is, it can be summed up as "sometimes the harder way is the easier way". For Chicago drivers, it was always 2 wrongs don't make a right, but 3 rights do make a left. Sykes and Siklossy were working together in the tower at the university of texas when a fire broke out on campus. A large crowd went to the top floor observation platform to watch the fire. When quitting time came, every time the down elevator doors opened, the cars were full from the people leaving the observation platform. After a couple of attempts, they pushed the up button. Got into an empty car and road it up to the observation platform, and then road down with the crowd. That started them thinking how to teach AI to think that way (1974 or 75, I don't remember exactly).
For interesting reading on what Sci-Fi had to say about such things long before the Terminator I suggest:
The Moon is a Harsh Mistress by Heinlein, AI computer joins the revolt on the moon and coordinates and basically runs the revolt Asimov's Robot Series, but most important is the 3rd book Robots of Dawn
My favorite movie on the subject Colossus: The Forbin Project
|
|
|
Post by Village Idiot on Oct 9, 2023 20:08:58 GMT -5
On a side note, if you are wondering what the insane heuristic is, it can be summed up as "sometimes the harder way is the easier way". For Chicago drivers, it was always 2 wrongs don't make a right, but 3 rights do make a left. Sykes and Siklossy were working together in the tower at the university of texas when a fire broke out on campus. A large crowd went to the top floor observation platform to watch the fire. When quitting time came, every time the down elevator doors opened, the cars were full from the people leaving the observation platform. After a couple of attempts, they pushed the up button. Got into an empty car and road it up to the observation platform, and then road down with the crowd. That started them thinking how to teach AI to think that way (1974 or 75, I don't remember exactly). Now that's interesting. That reminds me of a school custodian I know where no one could figure out how to reset all of the school clocks when daylight savings time ended. When everyone went home he shut the power off for an hour then started it again.
|
|
Dub
Administrator
I'm gettin' so the past is the only thing I can remember.
Posts: 19,916
|
Post by Dub on Oct 9, 2023 20:34:38 GMT -5
My favorite movie on the subject Colossus: The Forbin Project I went looking for that film earlier today to post a link in this thread but it isn't available for streaming anywhere. It was evidently on the Criterion Channel some time back but isn't there now.
|
|
|
Post by t-bob on Oct 9, 2023 21:00:58 GMT -5
My favorite movie on the subject Colossus: The Forbin Project I went looking for that film earlier today to post a link in this thread but it isn't available for streaming anywhere. It was evidently on the Criterion Channel some time back but isn't there now. go to the public library - the clerk will help you. If they can’t find it, you’ll have to find it on the Internet
|
|
|
Post by coachdoc on Dec 1, 2023 11:46:04 GMT -5
The Moon is a Harsh Mistress is one of my all time favorite stories.
|
|
|
Post by howard lee on Dec 1, 2023 12:48:06 GMT -5
|
|
|
Post by Cornflake on Dec 1, 2023 12:54:51 GMT -5
"To misquote Jeff Goldblum's character in Jurassic Park, They are so focused on the could we do this, they didn't bother to slow down and listen to those that asked, should we do this?"
One way this situation is different, I think, is that AI has potential military applications. Even if we were to decide not to do this, the Chinese would. It reminds me of the arms race in that regard. No one can afford to sit on the sidelines.
|
|
|
Post by dradtke on Dec 1, 2023 13:23:39 GMT -5
|
|
|
Post by Hobson on Dec 1, 2023 14:57:42 GMT -5
|
|
|
Post by PaulKay on Dec 2, 2023 9:40:10 GMT -5
In some ways I think the fear of AI getting out of control is just crazy hysteria over the unknown. At some point, it still takes a human to "give them" control over something.
|
|
|
Post by james on Dec 2, 2023 10:38:49 GMT -5
|
|