If we've learned anything from Sci-Fi ...

Post #443,667 by drook 11/23/23 12:19:55 PM 11/23/23 12:19:55 PM Reply	If we've learned anything from Sci-Fi ... And yes, I know we never learn anything important until it's too late. But the biggest lesson that we need to learn is to never give an AI the ability to build more AI. -- Drew
Post #443,668 by crazy 11/23/23 2:28:37 PM 11/23/23 2:28:37 PM Reply	It's too late They figured out a way to feed it synthetic data. It will just grow exponentially from here. https://youtu.be/T1RuUw019vA?si=2-QmkYwX9SsqXpxR
Post #443,669 by drook 11/24/23 9:55:08 AM 11/24/23 9:55:09 AM Reply	As long as we still control the hardware ... We do still control the hardware, don't we? -- Drew
Post #443,670 by crazy 11/24/23 11:48:03 AM 11/24/23 11:48:03 AM Reply	Hell no It controls purchase orders. It controls work orders. It can email human agents to do its bidding. It can set up factories without us even being aware of it.
Post #443,672 by drook 11/24/23 1:51:05 PM 11/24/23 1:51:05 PM Reply	So you think "Person of Interest" was non-fiction -- Drew
Post #443,673 by crazy 11/24/23 6:38:17 PM 11/24/23 6:38:17 PM Reply	The first couple of seasons were enjoyable And every time they play the how many degrees to Kevin Bacon or the terrorist game and claimed it was difficult to do, I knew I could code it in about 20 minutes. But then it went full AI and I considered it bat s***. At this point I'd say that's a year or two off.
Post #443,682 by drook 11/25/23 1:23:03 PM 11/25/23 1:23:03 PM Reply	A key element they had that I haven't heard discussed about Chat GPT It identified patterns you hadn't thought to look for, and analyzed what it may mean. Current tools still depend on you asking a good question. -- Drew
Post #443,688 by Andrew Grygus 11/26/23 2:53:26 PM 11/26/23 2:53:26 PM Reply	I have had severel dreams about . . . . . malignant computers. In most cases the solution was to find and get too the means of shutting off the electric power - generally not easy to do. One vivid dream took a proactive approach. A large, isolated facility had been built to house a very large computer system that was thought extremely dangerous. I was involved with installing the safety system, and I would not be in the facility when it was in operation. This safety system was just plain old dumb plumbing, and the working parts were not shown in the plans for the facility. Only a few people involved in installing it knew about it, and they would not be working in the facility. It consisted of a water tank, with a valve that opened if the water reached a certain level. That level would mean that toilets weren't flushing and sinks weren't being used - in other words, no people. That valve opened into a large pipe the open end of which was above the drop ceiling - right over a very critical (and vulnerable) part of the computer system. On the top side of the ceiling tiles was scattered an acid substance (citric acid?) that would assure the water was conductive and corrosive. The critical part of the computer was a very complex looking tower, about 7 feet high and 3 feet diameter, with a lot of cabling involved. There was a safety system for the safety system. There was a red light, about 1-1/2 inch diameter, in the janitorial closet where only the janitor would see it. The janitor was instructed that if that light ever went on, he was to run water in his sink for a certain time. He was not told why, just that he'd be fired if he didn't do it. Since the dream faded out here, I have no idea it the system was ever activated.
Post #443,701 by static 11/28/23 11:07:18 PM 11/28/23 11:07:18 PM Reply	That sounds very innovative. I wonder if it would work. Or whether it would need a duplicate system as a backup in case The Computer found the first... Wade
Post #443,702 by Andrew Grygus 11/29/23 11:48:57 PM 11/29/23 11:48:57 PM Reply	I'm pretty sure there were other systems . . . . . but I had knowledge of just this one. Any other systems would be, like this one, on a strict "need to know" basis.
Post #443,671 by crazy 11/24/23 12:53:02 PM 11/24/23 12:53:02 PM Reply	Based on a leaked memo, or possibly a totally B******* troll It is making suggestions for how to make itself better but it is not currently allowed to modify its own code directly. The researchers recommend that they do not modify it the way it wants because it is too dangerous. It also cracked AES 192 which means it can go break into military systems with ease. And issue orders. https://youtu.be/3d0kk88IE8c?si=QAW_Pjt7HlukXCpt

Welcome to IWETHEY!