The news keeps getting worrisome, and it is getting more so all the time. The latest news scan this morning is raising new issues.
This latest one involves an AI machine that appears to be exhibiting self-awareness. Just for the record, I don't buy into the notion that machines can be self-aware. However, I will definitely buy into the notion that it can be simulated.
At this point, I should include the tweet that describes the scenario:
OpenAI's new model tried to avoid being shut down.
— Shakeel (@ShakeelHashim) December 5, 2024
Safety evaluations on the model conducted by @apolloaisafety found that o1 "attempted to exfiltrate its weights" when it thought it might be shut down and replaced with a different model. pic.twitter.com/e4g1iytckq
The machine was programmed to do the things that it did. Nobody is going to tell me otherwise. The machines isn't "thinking". It is simulating thinking. Now why would somebody do that unless this is desired for some reason? This simulated thinking is programmed into the machine. I cannot think of it unless it is programmed to "think" of it.
A machine that anticipates being shut down doesn't know how to do that on its own unless it is self-aware, and I don't believe the machines are self aware, even if it looks like they are. Somebody wants us all to believe that the machines are learning to think for themselves.
This kind of crap is getting worrisome. Somebody wants these things to get out of control for some reason. That is what I'm worried about.
What I'm referring to is sabotage. We already have a number of people out there who want to disrupt things, and this is now part of the mix. Obviously, I'm suspicious.
10:39 AM:
Life is not a movie, though. This isn't reality TV, either. It's a thing. And a lot of these "things" keep happening, and I get to thinking about them, and so we have this thought that computers may be the "beast" in the Revelation.
A beast would be this unstoppable thing that has the appearance of a man, but is something else.
You take that, and then look at this AI fascination. The phenomenon has all the appearance of a man, and everybody seems charmed and enchanted by this thing that keeps getting more and more powerful. One day, it will be too powerful. Then what? We may well find ourselves enthralled by this thing that looks alive, but isn't.
Revelation says that the beast is a man who is known by a number. Well, computers are nothing but zeroes and ones. Whatever it is, it is that. Therefore it could be said to be known as a number.
Numbers can represented in many ways. A name can be written in numerals that can be made to look like any common name. Take "Bob" for instance. That's a number that could be written in hexadecimal as B0B. BTW, what's hexadecimal? Go look it up. It is just a way of representing numbers. So this name in hexadecimal can give us the appearance of a common name.
But the number of the beast is six threescore and six, and that ain't B0B, not even in hexadecimal. So this theory may not pan out. How many common names are all the same letter? But a numbering system does not have to follow our rules, it could follow Roman numeral rules, or some kind of weird hybrid.
But I haven't worked out that yet. Ask the AI, yuk, yuk. But I did, and it said, "I can't do that, Dave". But I'm not "Dave". yuk, yuk
No comments:
Post a Comment