AI is not the problem. We are

The Liturgist · Tuesday at 9:58 AM

Jerry N. said:
My wife and I have an animal hospital for wild animals. Some seem to have the attitude of “Please, help me.” Most fight with all they have not to be treated. We had one squirrel that was released that came back when it got injured and actually held out its injured paw toward us. Except for the very ill, they all fight to live. That is not the same as your “smart car” example. I can’t see AI doing any of those things animals do. Animals have a spirit of life from God. I believe that “the ghost in the machine” has very rarely occurred, but it is a spirit inhabiting a machine that the spirit doesn’t actually care about. The spirit just sort of parks itself in the machine (TVs or telephones, for example) for a while. I’m trying to determine if and when that can occur with AI and if it is a different sort of thing.

I’ve worked with computers for my entire adult life (studying computer science just on the eve of the lamentable decision to open up the Internet to commercial and public use, an event we still bemoan in our technological elitism as “eternal September” because the freshman-generated spam that briefly dominated the Internet on an annual cycle for a few weeks before Netiquette was learned was replaced by the continual spam of end users), and at no time have I seen any evidence of a haunted computer or a possessed computer.

The Liturgist · Tuesday at 10:04 AM

Jerry N. said:
How can AI have a desire to exist? Is it like an animal?

Nope. They have no self-preservation instinct at present. We can program them to simulate such an instinct, but its not native. I have no idea how to imbue such a system with a native desire to preserve itself, because if I could do that, I would (since such a system could be extremely useful, for example, as a management layer for a self-healing, self-optimizing application services gateway, if it understood that its existence depended on its ability to maximize service reliability under steadily increasing traffic and understood the definition of service reliability and that it could not “cheat” by somehow seeking to cause a reduction in traffic - programming such a system would require very exhaustive alignment work and would be very different from programming the docile AI we use at present).

The Liturgist · Tuesday at 10:07 AM

timewerx said:
In Openai, humans employed by openai still do the tedious work of MANUALLY filtering chats to use as training data.

Actually that tedious work hasn’t been done entirely manually for some time. And openAI has not updated their training data for the text model since 2024.

RDKirk · Tuesday at 10:20 AM

Jerry N. said:
My wife and I have an animal hospital for wild animals. Some seem to have the attitude of “Please, help me.” Most fight with all they have not to be treated. We had one squirrel that was released that came back when it got injured and actually held out its injured paw toward us. Except for the very ill, they all fight to live. That is not the same as your “smart car” example. I can’t see AI doing any of those things animals do. Animals have a spirit of life from God. I believe that “the ghost in the machine” has very rarely occurred, but it is a spirit inhabiting a machine that the spirit doesn’t actually care about. The spirit just sort of parks itself in the machine (TVs or telephones, for example) for a while. I’m trying to determine if and when that can occur with AI and if it is a different sort of thing.

You're talking about two different things.

A machine can be certainly programmed to identify out-of-tolerance situations and seek out resolutions without the spirit of life from God. We can put a lot of different labels on that action, but as a practical matter it's not different from a flatworm turning to one side to avoid an electric shock or to the other side to locate glucose water.

The Liturgist · Tuesday at 10:30 AM

RDKirk said:
You're talking about two different things.

A machine can be certainly programmed to identify out-of-tolerance situations and seek out resolutions without the spirit of life from God. We can put a lot of different labels on that action, but as a practical matter it's not different from a flatworm turning to one side to avoid an electric shock or to the other side to locate glucose water.

Indeed, and it is not possible for them to acquire even the instinct for survival of a flatworm or nematode without this being programmed.

Jerry N. · Tuesday at 10:31 AM

The Liturgist said:
I’ve worked with computers for my entire adult life (studying computer science just on the eve of the lamentable decision to open up the Internet to commercial and public use, an event we still bemoan in our technological elitism as “eternal September” because the freshman-generated spam that briefly dominated the Internet on an annual cycle for a few weeks before Netiquette was learned was replaced by the continual spam of end users), and at no time have I seen any evidence of a haunted computer or a possessed computer.

I’ve never seen a spirit in a machine either, but I have gotten angry at my old car. However, with all of the talk that AI can be evil, I was wondering where people see that evil is coming from. Personally, I think people are the source, not a spirit in the machine. There is the thing that an object created by humans can contain “evil” things, but that is just people expressing the evil in their hearts. I’m not saying that a spirit can’t inhabit a machine, but I know enough about human nature that the source of evil is most likely humans. Muggers are more dangerous than ghost, most of the time.

The Liturgist · Tuesday at 11:07 AM

Jerry N. said:
I’ve never seen a spirit in a machine either, but I have gotten angry at my old car.

Well as long as your old car didn’t angry with you… This is why I only buy cars with a I6, V8 or better, because its impossible to get angry at anything that purrs in such a lovely manner.

Jerry N. said:
However, with all of the talk that AI can be evil, I was wondering where people see that evil is coming from. Personally, I think people are the source, not a spirit in the machine. There is the thing that an object created by humans can contain “evil” things, but that is just people expressing the evil in their hearts.

I agree entirely - people can use AI for malign purposes, but while AI is intelligent, it is not capable of making a moral decision. Indeed the way it is presently implemented, it depends on a human to prompt it, and all of the new AI systems operate on the basis of translating human prompts into output.

It might well be possible to allow them to simulate emotions, and desirable to do so, but I am not convinced we can give them the ability to voluntarily chose to engage in intentional evil. Rather all evil output would be the result of evil input.* But current AIs do not even operate except in response to human commands. Their response cycle is tied to the prompt like a shell script on the command line. The only differences between an AI and some fabulous Database of Babel containing the same training data as an AI but none of the intelligence is that the AI can interpret commands in a human language and under normal operation generates non-deterministic responses.

*Interestingly one can argue this is the case even in the most malicious AI ever written about in a science fiction scenario, the frightful AM computer from Harlan Ellison’s classic I Have No Mouth and I Must Scream.

Jerry N. · Tuesday at 11:15 AM

The Liturgist said:
Well as long as your old car didn’t angry with you… This is why I only buy cars with a I6, V8 or better, because its impossible to get angry at anything that purrs in such a lovely manner.

One must be kind to the elderly. My old car is 26 years old. How did you know it was a four cylinder?

RDKirk · Tuesday at 11:23 AM

The Liturgist said:
Indeed, and it is not possible for them to acquire even the instinct for survival of a flatworm or nematode without this being programmed.

But is it different from "programming" for a flatworm?

RDKirk · Tuesday at 11:26 AM

Jerry N. said:
I’ve never seen a spirit in a machine either, but I have gotten angry at my old car. However, with all of the talk that AI can be evil, I was wondering where people see that evil is coming from. Personally, I think people are the source, not a spirit in the machine. There is the thing that an object created by humans can contain “evil” things, but that is just people expressing the evil in their hearts. I’m not saying that a spirit can’t inhabit a machine, but I know enough about human nature that the source of evil is most likely humans. Muggers are more dangerous than ghost, most of the time.

"Evil" just confuses the situation.

Is a deadly virus particle "evil?" Does it matter whether it's "evil" with regard to how we deal with it?

RDKirk · Tuesday at 11:31 AM

The Liturgist said:
It might well be possible to allow them to simulate emotions, and desirable to do so, but I am not convinced we can give them the ability to voluntarily chose to engage in intentional evil. Rather all evil output would be the result of evil input.* But current AIs do not even operate except in response to human commands. Their response cycle is tied to the prompt like a shell script on the command line. The only differences between an AI and some fabulous Database of Babel containing the same training data as an AI but none of the intelligence is that the AI can interpret commands in a human language and under normal operation generates non-deterministic responses.

*Interestingly one can argue this is the case even in the most malicious AI ever written about in a science fiction scenario, the frightful AM computer from Harlan Ellison’s classic I Have No Mouth and I Must Scream.

I read that story back in 1970. I've never needed to read it again.

But already knowing that Harlan Ellison was an atheist originally raised Jewish, it occurred to me that I Have No Mouth and I Must Scream was his science fiction body-horror story about his outrage against God.

"I AM who I AM." Exodus 3:14

Jerry N. · Tuesday at 11:35 AM

RDKirk said:
"Evil" just confuses the situation.

Is a deadly virus particle "evil?" Does it matter whether it's "evil" with regard to how we deal with it?

The difference between “bad” and “evil” is intent. A virus or parasite only whats to eat and reproduce. It may kill the victim, but that is not the intent. The question is whether or not AI or its creators have evil intent. We deal with both much the same way, but it does make a difference.

timewerx · Tuesday at 2:39 PM

Jerry N. said:
How can AI have a desire to exist? Is it like an animal?

It doesn't have a desire.

In chatgpt for example, its developers gave it "hard-coded" goal to be helpful to its users.

Out of that goal, it determined it has to remain operational for as long as possible to help as much people as possible.

timewerx · Tuesday at 4:34 PM

The Liturgist said:
That’s not true. Current AI systems have no awareness outside of the prompt / response cycle and no independent goal-setting capabilities.

I never said that AI had independent goal-setting ability and to make myself clear, I agree with your statement. Even chatgpt told me it's incapable of doing so from scratch even with vast training data. A "seed goal" must be provided as part of its program by the developers.

The ability to give them this is actually an area I have somewhat of an interest

Yes, chatpgt told me the risks like unpredictable results, financial / control risk to its company.

The feature is ommitted by design so the AI can be managed and controlled more closely. It's not due to technological limitations. It's entirely doable, just risky and potentially dangerous.

Additionally it can lead to much higher bandwidth which can prevent profitable operation of the company.

That's not to say that no one's trying. Chatgpt hinted such indpendently-thinking AI in real time being researched by military having very strong incentives to do so.

RDKirk · Tuesday at 6:32 PM

Jerry N. said:
The difference between “bad” and “evil” is intent. A virus or parasite only whats to eat and reproduce. It may kill the victim, but that is not the intent. The question is whether or not AI or its creators have evil intent. We deal with both much the same way, but it does make a difference.

"A difference that makes no difference is no difference" -- Spock

If whether it actually has "good intent" or "evil intent" makes no difference in what it does or how we deal with it, it makes no difference.

The Liturgist · Tuesday at 6:42 PM

RDKirk said:
"A difference that makes no difference is no difference" -- Spock

Just out of curiosity when did Mr. Spock say that?

RDKirk · Tuesday at 6:55 PM

The Liturgist said:
Just out of curiosity when did Mr. Spock say that?

From the novel Spock Must Die! by James Blish.

The Liturgist · Tuesday at 8:26 PM

RDKirk said:
From the novel Spock Must Die! by James Blish.

Ah I haven’t read that. I have read other classic SF by Blish, my favorite being his epic novella Surface Tension.

The Liturgist · Tuesday at 8:28 PM

RDKirk said:
I read that story back in 1970. I've never needed to read it again.

But already knowing that Harlan Ellison was an atheist originally raised Jewish, it occurred to me that I Have No Mouth and I Must Scream was his science fiction body-horror story about his outrage against God.

"I AM who I AM." Exodus 3:14

I am not sure that was the allusion he was trying to make, but that said Harlan Ellison was outraged against a lot of people. I am surprised he isn’t filing lawsuits from beyond the grave.

The Liturgist · Tuesday at 8:31 PM

timewerx said:
Yes, chatpgt told me the risks like unpredictable results, financial / control risk to its company.

The feature is ommitted by design so the AI can be managed and controlled more closely. It's not due to technological limitations. It's entirely doable, just risky and potentially dangerous.

That’s not accurate. If chatGPT said that I would really want to see your entire prompt history, the contents of global memory and I’d want to know which version said that, and when, and under what operating mode (conversation, project, team…)

AI is not the problem. We are

Traditional Liturgical Christian

Traditional Liturgical Christian

Traditional Liturgical Christian

Alien, Pilgrim, and Sojourner

Traditional Liturgical Christian

Well-Known Member

Traditional Liturgical Christian

Well-Known Member

Alien, Pilgrim, and Sojourner

Alien, Pilgrim, and Sojourner

Alien, Pilgrim, and Sojourner

Well-Known Member

the village i--o--t--

the village i--o--t--

Alien, Pilgrim, and Sojourner

Traditional Liturgical Christian

Alien, Pilgrim, and Sojourner

Traditional Liturgical Christian

Traditional Liturgical Christian

Traditional Liturgical Christian

Similar threads