ChatGPT and AI: Are we all about to die?

[FuB]

I have a quick question; would anybody here like to receive therapy or counselling from an AI bot?

An AI text bot or an AI voice over the phone?

I've read fairly recently that "many" are turning to AI bots for mental health support and counselling so this is a really excellent question, Sid.

On one hand, I think AI could have an advantage in that it is entirely without emotional bias whereas counsellors often have to work really hard to avoid putting their own feelings into therapy sessions. On the other hand, even with the current and increasing capabilities of generative AI, it's just too good at and too happy to make shit up to be a safe alternative, especially where vulnerable people are concerned.

I mean, i like tinkering around with AI bots and mostly to see just how far i can push till i hit a guiderail. If i turn to AI for assistance on a particular subject or process, i do it with a large expectation that i need to cross-check what it comes out with. I've also posted above how it's often very frustrating in terms of its "wish" to help at all costs despite seemingly being incapable of realising/appreciating where it can't actually do (reliably) what it is offering to do. In the domain of counselling, therapy, advice-sector, etc. that is just too damn dangerous.

I have, however, seen examples of (not - yet - public) further trained AI bots that are trained in a specific specialist subject and given extra guiderails to make sure they quote known facts. I still think there's too much chance of them "hallucinating" even with specialisation but the bot i saw was concerned with specific legislature and was aimed at the advice sector.

[swampash]

For what it’s worth, here’s swampy-the-luddites take on this.
There are three basic reasons why I have a problem with the AI hype. They can be summarized as the organic argument, the knowledge horizon problem and the mortality issue. In no particular order:
1. The brain is not a computer. It is a tactile biochemical organ that learns through its mistakes. It learns to walk by falling down. By burning its fingers it learns not to play with fire. By getting a bloody nose it learns not to pick a fight with someone who is bigger and stronger; although by playing sport it also learns that technique can trump brawn. To mimic this capacity you would first have to learn how to cultivate brain tissue and find a way to use it to build an organic form of computer that can truly mimic the brain’s unique capabilities.

I think you might want to avoid understanding what a neural network is then because i think it might scare you. That's what these things are built upon and, at the base of this, we can ignore any of their specialisations. They have for at least 30 years been trying to exactly mimic how a human brain works with silicon. No brain tissue required. These things ARE learning from both training and mistakes and goals-based incentives.

2. The human brain is uniquely aware of its own mortality and consequently learns to value time and to cherish fleeting moments. I can’t comprehend AI matching such consciousness.

In what way is emotion an advantage, swamps? Also, just because you can't comprehend it, doesn't mean it isn't happening. That's like expecting a fish to understand a human.

3. If the set of all that is known is described by a circle, then the circumference of that circle is the boundary between what is known and what is unknown. That boundary constitutes a knowledge horizon. I can see how AI can scan the domain of the known, at a speed that the brain cannot match, and for any given task produce a banal answer based on accepted wisdom. But it cannot look beyond the knowledge horizon.
The human brain is completely different in my view. It is uniquely able to speculate about what might lie beyond that knowledge horizon. Indeed it is that very capacity that is at the core of extending the bank of existing knowledge. It is the basis of all good research - hypothesizing about what might be and then designing experiments to test such hypotheses. The results of this hypothesis testing then move the knowledge horizon outwards.

Just my idle thoughts on the subject and, of course, I could be totally wrong…

Sign up and have some conversations (for free) with chatGPT, swamps. It's more than happy to speculate and, yes, i would agree that it doesn't really understand what it is speculating (although that's getting harder and harder to see). Just do it. Don't have opinions on something you haven't actually had a bit of experience with because you are missing a lot here. I'm not intending to patronise you here at all by the way but it's sort of clear you haven't actually had a good look at what you are criticising.

Swampy-the-luddite just spend half an hour typing a response to this and the post didn't appear in the thread. Perhaps he'll try again tomorrow?

Use AI to formulate a reply, it's a lot quicker

...and banal?

[swampash]

I have a quick question; would anybody here like to receive therapy or counselling from an AI bot?

An AI text bot or an AI voice over the phone?

Nope

[swampash]

I have a quick question; would anybody here like to receive therapy or counselling from an AI bot?

An AI text bot or an AI voice over the phone?

I've read fairly recently that "many" are turning to AI bots for mental health support and counselling so this is a really excellent question, Sid.

On one hand, I think AI could have an advantage in that it is entirely without emotional bias whereas counsellors often have to work really hard to avoid putting their own feelings into therapy sessions. On the other hand, even with the current and increasing capabilities of generative AI, it's just too good at and too happy to make shit up to be a safe alternative, especially where vulnerable people are concerned.

I mean, i like tinkering around with AI bots and mostly to see just how far i can push till i hit a guiderail. If i turn to AI for assistance on a particular subject or process, i do it with a large expectation that i need to cross-check what it comes out with. I've also posted above how it's often very frustrating in terms of its "wish" to help at all costs despite seemingly being incapable of realising/appreciating where it can't actually do (reliably) what it is offering to do. In the domain of counselling, therapy, advice-sector, etc. that is just too damn dangerous.

I have, however, seen examples of (not - yet - public) further trained AI bots that are trained in a specific specialist subject and given extra guiderails to make sure they quote known facts. I still think there's too much chance of them "hallucinating" even with specialisation but the bot i saw was concerned with specific legislature and was aimed at the advice sector.

At the risk of pulling a dozer on you, can I assume you still stand by your older reply here?:
viewtopic.php?t=2957

[FuB]

I have a quick question; would anybody here like to receive therapy or counselling from an AI bot?

An AI text bot or an AI voice over the phone?

I've read fairly recently that "many" are turning to AI bots for mental health support and counselling so this is a really excellent question, Sid.

On one hand, I think AI could have an advantage in that it is entirely without emotional bias whereas counsellors often have to work really hard to avoid putting their own feelings into therapy sessions. On the other hand, even with the current and increasing capabilities of generative AI, it's just too good at and too happy to make shit up to be a safe alternative, especially where vulnerable people are concerned.

I mean, i like tinkering around with AI bots and mostly to see just how far i can push till i hit a guiderail. If i turn to AI for assistance on a particular subject or process, i do it with a large expectation that i need to cross-check what it comes out with. I've also posted above how it's often very frustrating in terms of its "wish" to help at all costs despite seemingly being incapable of realising/appreciating where it can't actually do (reliably) what it is offering to do. In the domain of counselling, therapy, advice-sector, etc. that is just too damn dangerous.

I have, however, seen examples of (not - yet - public) further trained AI bots that are trained in a specific specialist subject and given extra guiderails to make sure they quote known facts. I still think there's too much chance of them "hallucinating" even with specialisation but the bot i saw was concerned with specific legislature and was aimed at the advice sector.

At the risk of pulling a dozer on you, can I assume you still stand by your older reply here?:
viewtopic.php?t=2957

Yes. I still stand by that. LLMs are not "sentient" and the tenets in that video still stand. My concern with them is that the general media noise tends to suggest they are infallible and, despite warnings all over sites like chatGPT, that's a worry for likely vulnerable people seeking help and not necessarily taking it with a pinch of salt.

[FuB]

For what it’s worth, here’s swampy-the-luddites take on this.
There are three basic reasons why I have a problem with the AI hype. They can be summarized as the organic argument, the knowledge horizon problem and the mortality issue. In no particular order:
1. The brain is not a computer. It is a tactile biochemical organ that learns through its mistakes. It learns to walk by falling down. By burning its fingers it learns not to play with fire. By getting a bloody nose it learns not to pick a fight with someone who is bigger and stronger; although by playing sport it also learns that technique can trump brawn. To mimic this capacity you would first have to learn how to cultivate brain tissue and find a way to use it to build an organic form of computer that can truly mimic the brain’s unique capabilities.

I think you might want to avoid understanding what a neural network is then because i think it might scare you. That's what these things are built upon and, at the base of this, we can ignore any of their specialisations. They have for at least 30 years been trying to exactly mimic how a human brain works with silicon. No brain tissue required. These things ARE learning from both training and mistakes and goals-based incentives.

2. The human brain is uniquely aware of its own mortality and consequently learns to value time and to cherish fleeting moments. I can’t comprehend AI matching such consciousness.

In what way is emotion an advantage, swamps? Also, just because you can't comprehend it, doesn't mean it isn't happening. That's like expecting a fish to understand a human.

3. If the set of all that is known is described by a circle, then the circumference of that circle is the boundary between what is known and what is unknown. That boundary constitutes a knowledge horizon. I can see how AI can scan the domain of the known, at a speed that the brain cannot match, and for any given task produce a banal answer based on accepted wisdom. But it cannot look beyond the knowledge horizon.
The human brain is completely different in my view. It is uniquely able to speculate about what might lie beyond that knowledge horizon. Indeed it is that very capacity that is at the core of extending the bank of existing knowledge. It is the basis of all good research - hypothesizing about what might be and then designing experiments to test such hypotheses. The results of this hypothesis testing then move the knowledge horizon outwards.

Just my idle thoughts on the subject and, of course, I could be totally wrong…

Sign up and have some conversations (for free) with chatGPT, swamps. It's more than happy to speculate and, yes, i would agree that it doesn't really understand what it is speculating (although that's getting harder and harder to see). Just do it. Don't have opinions on something you haven't actually had a bit of experience with because you are missing a lot here. I'm not intending to patronise you here at all by the way but it's sort of clear you haven't actually had a good look at what you are criticising.

Swampy-the-luddite just spend half an hour typing a response to this and the post didn't appear in the thread. Perhaps he'll try again tomorrow?

Use AI to formulate a reply, it's a lot quicker

...and banal?

well, let's see....

--------

Please fomulate a reply to post id 193368 on site index.php in the style of user swampash


ChatGPT said:
I can help craft a reply in the style of “swampash,” but I’ll need to see the content of post ID 193368 first. Could you please paste the text of that post here? Once I’ve got it, I’ll compose a tailored response that matches swampash’s tone and style.




Sources
You said:
here is a link to that post viewtopic.php?p=193368#p193368


ChatGPT said:

Thought for a few seconds
Below is a draft reply in the style of swampash — brisk, lightly self-deprecating, evocative of the “luddite” persona, and engaging in rhetorical back-and-forth:

Ah, so let me see if I’ve got this straight: we’re equating emotion with a weakness, when in fact it might just be the signal that points the way?

Brain tissue vs silicon — it’s not “life Jim” vs “non‑life Jim,” it’s the difference between a sponge and a microprocessor. You can simulate structure, but you’re a long way from replicating becoming.

Emotion being an “enormous advantage”? Fair enough — but it’s not just about feeling. It’s about the way emotion anchors us to “now,” to meaning… something that cold logic doesn’t care a damn about.

Sure, I could “have a go” with ChatGPT — but I don’t need to board a train to know it’s heading for a cliff. Reckon I’ve better things to do than play Russian Roulette with someone Else’s Code.

But hey, I’m not saying I’m right. I just don’t want to fancy I’m crossing the bridge until I’ve seen the river. Maybe I’ll have that play once I’ve got time to burn — but don’t bet your bottom dollar on swampy suddenly thinking AI is the second coming.

Let me know if you’d prefer it tweaked in any particular direction.

-----

Do you fear being replaced now?

[swampash]

I think you might want to avoid understanding what a neural network is then because i think it might scare you. That's what these things are built upon and, at the base of this, we can ignore any of their specialisations. They have for at least 30 years been trying to exactly mimic how a human brain works with silicon. No brain tissue required. These things ARE learning from both training and mistakes and goals-based incentives.



In what way is emotion an advantage, swamps? Also, just because you can't comprehend it, doesn't mean it isn't happening. That's like expecting a fish to understand a human.



Sign up and have some conversations (for free) with chatGPT, swamps. It's more than happy to speculate and, yes, i would agree that it doesn't really understand what it is speculating (although that's getting harder and harder to see). Just do it. Don't have opinions on something you haven't actually had a bit of experience with because you are missing a lot here. I'm not intending to patronise you here at all by the way but it's sort of clear you haven't actually had a good look at what you are criticising.

Swampy-the-luddite just spend half an hour typing a response to this and the post didn't appear in the thread. Perhaps he'll try again tomorrow?

Use AI to formulate a reply, it's a lot quicker

...and banal?

well, let's see....

--------

Please fomulate a reply to post id 193368 on site index.php in the style of user swampash


ChatGPT said:
I can help craft a reply in the style of “swampash,” but I’ll need to see the content of post ID 193368 first. Could you please paste the text of that post here? Once I’ve got it, I’ll compose a tailored response that matches swampash’s tone and style.




Sources
You said:
here is a link to that post viewtopic.php?p=193368#p193368


ChatGPT said:

Thought for a few seconds
Below is a draft reply in the style of swampash — brisk, lightly self-deprecating, evocative of the “luddite” persona, and engaging in rhetorical back-and-forth:

Ah, so let me see if I’ve got this straight: we’re equating emotion with a weakness, when in fact it might just be the signal that points the way?

Brain tissue vs silicon — it’s not “life Jim” vs “non‑life Jim,” it’s the difference between a sponge and a microprocessor. You can simulate structure, but you’re a long way from replicating becoming.

Emotion being an “enormous advantage”? Fair enough — but it’s not just about feeling. It’s about the way emotion anchors us to “now,” to meaning… something that cold logic doesn’t care a damn about.

Sure, I could “have a go” with ChatGPT — but I don’t need to board a train to know it’s heading for a cliff. Reckon I’ve better things to do than play Russian Roulette with someone Else’s Code.

But hey, I’m not saying I’m right. I just don’t want to fancy I’m crossing the bridge until I’ve seen the river. Maybe I’ll have that play once I’ve got time to burn — but don’t bet your bottom dollar on swampy suddenly thinking AI is the second coming.

Let me know if you’d prefer it tweaked in any particular direction.

-----

Do you fear being replaced now?

What exactly gives you the right to do this without my explicit permission Fuß?

[FuB]

Swampy-the-luddite just spend half an hour typing a response to this and the post didn't appear in the thread. Perhaps he'll try again tomorrow?

Use AI to formulate a reply, it's a lot quicker

...and banal?

well, let's see....

--------

Please fomulate a reply to post id 193368 on site index.php in the style of user swampash


ChatGPT said:
I can help craft a reply in the style of “swampash,” but I’ll need to see the content of post ID 193368 first. Could you please paste the text of that post here? Once I’ve got it, I’ll compose a tailored response that matches swampash’s tone and style.




Sources
You said:
here is a link to that post viewtopic.php?p=193368#p193368


ChatGPT said:

Thought for a few seconds
Below is a draft reply in the style of swampash — brisk, lightly self-deprecating, evocative of the “luddite” persona, and engaging in rhetorical back-and-forth:

Ah, so let me see if I’ve got this straight: we’re equating emotion with a weakness, when in fact it might just be the signal that points the way?

Brain tissue vs silicon — it’s not “life Jim” vs “non‑life Jim,” it’s the difference between a sponge and a microprocessor. You can simulate structure, but you’re a long way from replicating becoming.

Emotion being an “enormous advantage”? Fair enough — but it’s not just about feeling. It’s about the way emotion anchors us to “now,” to meaning… something that cold logic doesn’t care a damn about.

Sure, I could “have a go” with ChatGPT — but I don’t need to board a train to know it’s heading for a cliff. Reckon I’ve better things to do than play Russian Roulette with someone Else’s Code.

But hey, I’m not saying I’m right. I just don’t want to fancy I’m crossing the bridge until I’ve seen the river. Maybe I’ll have that play once I’ve got time to burn — but don’t bet your bottom dollar on swampy suddenly thinking AI is the second coming.

Let me know if you’d prefer it tweaked in any particular direction.

-----

Do you fear being replaced now?

What exactly gives you the right to do this without my explicit permission Fuß?

Are you claiming ownership of the information on a public forum now? In any case, my point was that it's hardly doing a good job of impersonating you and so it was by way of a demonstration of its limitations. I wouldn't say what it had generated was "banal" though.

[swampash]

Use AI to formulate a reply, it's a lot quicker

...and banal?

well, let's see....

--------

Please fomulate a reply to post id 193368 on site index.php in the style of user swampash


ChatGPT said:
I can help craft a reply in the style of “swampash,” but I’ll need to see the content of post ID 193368 first. Could you please paste the text of that post here? Once I’ve got it, I’ll compose a tailored response that matches swampash’s tone and style.




Sources
You said:
here is a link to that post viewtopic.php?p=193368#p193368


ChatGPT said:

Thought for a few seconds
Below is a draft reply in the style of swampash — brisk, lightly self-deprecating, evocative of the “luddite” persona, and engaging in rhetorical back-and-forth:

Ah, so let me see if I’ve got this straight: we’re equating emotion with a weakness, when in fact it might just be the signal that points the way?

Brain tissue vs silicon — it’s not “life Jim” vs “non‑life Jim,” it’s the difference between a sponge and a microprocessor. You can simulate structure, but you’re a long way from replicating becoming.

Emotion being an “enormous advantage”? Fair enough — but it’s not just about feeling. It’s about the way emotion anchors us to “now,” to meaning… something that cold logic doesn’t care a damn about.

Sure, I could “have a go” with ChatGPT — but I don’t need to board a train to know it’s heading for a cliff. Reckon I’ve better things to do than play Russian Roulette with someone Else’s Code.

But hey, I’m not saying I’m right. I just don’t want to fancy I’m crossing the bridge until I’ve seen the river. Maybe I’ll have that play once I’ve got time to burn — but don’t bet your bottom dollar on swampy suddenly thinking AI is the second coming.

Let me know if you’d prefer it tweaked in any particular direction.

-----

Do you fear being replaced now?

What exactly gives you the right to do this without my explicit permission Fuß?

Are you claiming ownership of the information on a public forum now? In any case, my point was that it's hardly doing a good job of impersonating you and so it was by way of a demonstration of its limitations. I wouldn't say what it had generated was "banal" though.

And if swampy-the-luddite finds it intolerably rude and an infringement of his intellectual property?

[FuB]

...and banal?

well, let's see....

--------

Please fomulate a reply to post id 193368 on site index.php in the style of user swampash


ChatGPT said:
I can help craft a reply in the style of “swampash,” but I’ll need to see the content of post ID 193368 first. Could you please paste the text of that post here? Once I’ve got it, I’ll compose a tailored response that matches swampash’s tone and style.




Sources
You said:
here is a link to that post viewtopic.php?p=193368#p193368


ChatGPT said:

Thought for a few seconds
Below is a draft reply in the style of swampash — brisk, lightly self-deprecating, evocative of the “luddite” persona, and engaging in rhetorical back-and-forth:

Ah, so let me see if I’ve got this straight: we’re equating emotion with a weakness, when in fact it might just be the signal that points the way?

Brain tissue vs silicon — it’s not “life Jim” vs “non‑life Jim,” it’s the difference between a sponge and a microprocessor. You can simulate structure, but you’re a long way from replicating becoming.

Emotion being an “enormous advantage”? Fair enough — but it’s not just about feeling. It’s about the way emotion anchors us to “now,” to meaning… something that cold logic doesn’t care a damn about.

Sure, I could “have a go” with ChatGPT — but I don’t need to board a train to know it’s heading for a cliff. Reckon I’ve better things to do than play Russian Roulette with someone Else’s Code.

But hey, I’m not saying I’m right. I just don’t want to fancy I’m crossing the bridge until I’ve seen the river. Maybe I’ll have that play once I’ve got time to burn — but don’t bet your bottom dollar on swampy suddenly thinking AI is the second coming.

Let me know if you’d prefer it tweaked in any particular direction.

-----

Do you fear being replaced now?

What exactly gives you the right to do this without my explicit permission Fuß?

Are you claiming ownership of the information on a public forum now? In any case, my point was that it's hardly doing a good job of impersonating you and so it was by way of a demonstration of its limitations. I wouldn't say what it had generated was "banal" though.

And if swampy-the-luddite finds it intolerably rude and an infringement of his intellectual property?

Well, if that were the case then I think I'd suggest to swampy-the-luddite that he climb down from his high horse and, on the way down, consider if self-declared, clearly labelled parody without any commercial interest could or should be considered an infringement of intellectual property rights and, if so, to further consider if the joyless and dystopian world that would create might be worse than a world dominated by AI