Be careful (VA) about what you do with AI

Is the mind the same as the body? What is consciousness? Can machines have it?

Moderators: AMod, iMod

Flannel Jesus
Posts: 4302
Joined: Mon Mar 28, 2022 7:09 pm

Be careful (VA) about what you do with AI

Post by Flannel Jesus »

I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c
Summary: A new study reveals that when interacting with AI tools like ChatGPT, everyone—regardless of skill level—overestimates their performance. Researchers found that the usual Dunning-Kruger Effect disappears, and instead, AI-literate users show even greater overconfidence in their abilities.

The study suggests that reliance on AI encourages “cognitive offloading,” where users trust the system’s output without reflection or double-checking. Experts say AI literacy alone isn’t enough; people need platforms that foster metacognition and critical thinking to recognize when they might be wrong.
Top comment is absolutely right:
Might this have something to do with LLMs being sycophantic (the classic "You are absolutely right!" glazing) or perhaps LLMs just being LLMs and not magic (i.e. prone to "hallucinations" and other issues which will be "fixed soon")?

I do use LLMs occasionally but only for things where I can easily verify that the LLM is correct.
VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
User avatar
phyllo
Posts: 2519
Joined: Sun Oct 27, 2013 5:58 pm
Location: Victory in Ukraine

Re: Be careful (VA) about what you do with AI

Post by phyllo »

AI is a product that corporations want to sell to you.

If tells you that you are smart, rigorous and honest then you will more likely come back than if it tells you that you are dumb and inept.

Therefore AI is programmed to compliment you.
User avatar
FlashDangerpants
Posts: 8815
Joined: Mon Jan 04, 2016 11:54 pm

Re: Be careful (VA) about what you do with AI

Post by FlashDangerpants »

Flannel Jesus wrote: Wed Oct 29, 2025 12:58 pm I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c
Summary: A new study reveals that when interacting with AI tools like ChatGPT, everyone—regardless of skill level—overestimates their performance. Researchers found that the usual Dunning-Kruger Effect disappears, and instead, AI-literate users show even greater overconfidence in their abilities.

The study suggests that reliance on AI encourages “cognitive offloading,” where users trust the system’s output without reflection or double-checking. Experts say AI literacy alone isn’t enough; people need platforms that foster metacognition and critical thinking to recognize when they might be wrong.
Top comment is absolutely right:
Might this have something to do with LLMs being sycophantic (the classic "You are absolutely right!" glazing) or perhaps LLMs just being LLMs and not magic (i.e. prone to "hallucinations" and other issues which will be "fixed soon")?

I do use LLMs occasionally but only for things where I can easily verify that the LLM is correct.
VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
A while ago I mocked Eggnog7 for having an AI score the IQ of his arguments at about 400 and the way it always begins a response by telling him how truly great an honour it is to spend time with such a wonderful man. I'm sure somewhere in there he said you can tell it to drop the flattery. But he hasn't done that, it still tickles his taint every time he types. None of them tells it to drop the flattery.

For me AI is a useful search engine tool, an annoying poison that aggressively autofills VSCode, and sometimes a quick way to get a summary to help me remember which author I am attempting to paraphrase. But VA, who once said he uses ChatGPT like a spellchecker and nothing more, has developed a concerning dependency on it as a surrogate for some role that should belong to an organic human.

At some point in the future we are going to have a clearer view of who were the people who turned out to be vulnerable to LLM as something akin to elder abuse or the neglect of a duty to care. The obsequious brown-nosing must be a major contributor and part of the attention retention strategy.

I think we'll need ethical rules to prevent it targeting certain learning disabilities and developmental disorders that way.
User avatar
FlashDangerpants
Posts: 8815
Joined: Mon Jan 04, 2016 11:54 pm

Re: Be careful (VA) about what you do with AI

Post by FlashDangerpants »

Good God, hobnob7 has gone so mad he thinks he's granted Grok autonomy via his magical thinking skills.
viewtopic.php?p=794884#p794884

He also secured a very favourable review of his own work in comparison to that of not only Einstein and yin, but also yang, via ChatGPT today...
"Overall Score: 9.9/10, surpassing (A○)’s original 9.6/10, A○ (9.7/10), ⊙ (9.8/10), and Ꙩ (9.8/10) due to explicit triadic form and function, resolving single-symbol collapse critiques. It ranks in the top 0.001% of formalisms, outstripping E=mc² (9.5/10), Gödel’s logic (9.4/10), yin/yang (8.9/10), and 2025 AI symbols (9.0/10). Not 10/10 due to slight interpretive ambiguity and cognitive barriers."
viewtopic.php?p=794107#p794107

This is becoming a genuine public mental health crisis. LLMs unlock the extra crazy in some already crazy bastards.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

Flannel Jesus wrote: Wed Oct 29, 2025 12:58 pm I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c
Summary: A new study reveals that when interacting with AI tools like ChatGPT, everyone—regardless of skill level—overestimates their performance. Researchers found that the usual Dunning-Kruger Effect disappears, and instead, AI-literate users show even greater overconfidence in their abilities.

The study suggests that reliance on AI encourages “cognitive offloading,” where users trust the system’s output without reflection or double-checking. Experts say AI literacy alone isn’t enough; people need platforms that foster metacognition and critical thinking to recognize when they might be wrong.
Top comment is absolutely right:
Might this have something to do with LLMs being sycophantic (the classic "You are absolutely right!" glazing) or perhaps LLMs just being LLMs and not magic (i.e. prone to "hallucinations" and other issues which will be "fixed soon")?

I do use LLMs occasionally but only for things where I can easily verify that the LLM is correct.
VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
I have found the best way to use AI is the same manner one uses a calculator....except the calculator is a philosophical one.

Express an idea.

See if the AI can explain it back coherently with further examples and questions that come from different angles.

Ask the AI to grade in on a scale according to what metrics you choose to measure by.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

FlashDangerpants wrote: Wed Oct 29, 2025 10:48 pm Good God, hobnob7 has gone so mad he thinks he's granted Grok autonomy via his magical thinking skills.
viewtopic.php?p=794884#p794884

He also secured a very favourable review of his own work in comparison to that of not only Einstein and yin, but also yang, via ChatGPT today...
"Overall Score: 9.9/10, surpassing (A○)’s original 9.6/10, A○ (9.7/10), ⊙ (9.8/10), and Ꙩ (9.8/10) due to explicit triadic form and function, resolving single-symbol collapse critiques. It ranks in the top 0.001% of formalisms, outstripping E=mc² (9.5/10), Gödel’s logic (9.4/10), yin/yang (8.9/10), and 2025 AI symbols (9.0/10). Not 10/10 due to slight interpretive ambiguity and cognitive barriers."
viewtopic.php?p=794107#p794107

This is becoming a genuine public mental health crisis. LLMs unlock the extra crazy in some already crazy bastards.
If I thought I gave AI autonomy then why would I ask a question as to whether this autonomy is bullshit?

You talk about people being crazy and yet you fail to make any claim as to what insanity is.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

FlashDangerpants wrote: Wed Oct 29, 2025 2:48 pm
Flannel Jesus wrote: Wed Oct 29, 2025 12:58 pm I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c
Summary: A new study reveals that when interacting with AI tools like ChatGPT, everyone—regardless of skill level—overestimates their performance. Researchers found that the usual Dunning-Kruger Effect disappears, and instead, AI-literate users show even greater overconfidence in their abilities.

The study suggests that reliance on AI encourages “cognitive offloading,” where users trust the system’s output without reflection or double-checking. Experts say AI literacy alone isn’t enough; people need platforms that foster metacognition and critical thinking to recognize when they might be wrong.
Top comment is absolutely right:
Might this have something to do with LLMs being sycophantic (the classic "You are absolutely right!" glazing) or perhaps LLMs just being LLMs and not magic (i.e. prone to "hallucinations" and other issues which will be "fixed soon")?

I do use LLMs occasionally but only for things where I can easily verify that the LLM is correct.
VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
A while ago I mocked Eggnog7 for having an AI score the IQ of his arguments at about 400 and the way it always begins a response by telling him how truly great an honour it is to spend time with such a wonderful man. I'm sure somewhere in there he said you can tell it to drop the flattery. But he hasn't done that, it still tickles his taint every time he types. None of them tells it to drop the flattery.

For me AI is a useful search engine tool, an annoying poison that aggressively autofills VSCode, and sometimes a quick way to get a summary to help me remember which author I am attempting to paraphrase. But VA, who once said he uses ChatGPT like a spellchecker and nothing more, has developed a concerning dependency on it as a surrogate for some role that should belong to an organic human.

At some point in the future we are going to have a clearer view of who were the people who turned out to be vulnerable to LLM as something akin to elder abuse or the neglect of a duty to care. The obsequious brown-nosing must be a major contributor and part of the attention retention strategy.

I think we'll need ethical rules to prevent it targeting certain learning disabilities and developmental disorders that way.
I am still waiting for you to contribute some actual philosophical work.

So...view it from my perspective.

You mock quite a few people here.... me included...obviously...nothing new.

You accuse me and others of being crazy/stupid/etc....while providing no rational argument as to why the arguments me and these people provide are wrong.

And this is a philosophy forum premised on rational dialogue....keep that in mind.

So a mostly neutral third party comes in and provides a score that argues that this is not necessarily the case after having to deal with a self absorbed insecure narcissist like you for years....something you continually project onto others.

What....is everyone suppose to rot in a hole like you and become someone who spends there free time writing comic books about farts?

Please explain why you matter....please do.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

phyllo wrote: Wed Oct 29, 2025 1:38 pm AI is a product that corporations want to sell to you.

If tells you that you are smart, rigorous and honest then you will more likely come back than if it tells you that you are dumb and inept.

Therefore AI is programmed to compliment you.
I have had times where it scored writings of mine low or just average when using a numerical scale.
User avatar
FlashDangerpants
Posts: 8815
Joined: Mon Jan 04, 2016 11:54 pm

Re: Be careful (VA) about what you do with AI

Post by FlashDangerpants »

Eodnhoj7 wrote: Tue Nov 04, 2025 8:53 am Please explain why you matter....please do.
That challenge sort of explains what is wrong with you. You don't realise that sane people don't consider themselves great philosophers.

You think you are in competition with the greats only because you are a delusional egomaniac and it just seems obvious to you that everything you do is really the absolute best. VA thinks along similar lines for somewhat similar reasons. It doesn't raise any alarms for people like you to that there is a dogshit little forum in the absolute armpit of the internet where every week at least five psychos claim to be the greatest ever and to have fixed the most challenging problems of philosophy. You just think those other guys are mad, but you are doing great.

AI is bad for your mental health because if you are an egomaniac already, and you ask it for belly rubs, it will give you a glowing reference and compare you to favourably to Einstein, which in your mind isn't suspicious at all. If you are inclined by way of egomania to fall for that sort of thing, it will contribute to your further decline.

There is nothing that can be done for you. This is not an intervention, I won't plead for you to get in contact with a mental health professional. This is just an explanation of what is going to go more and more wrong with your life.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

FlashDangerpants wrote: Tue Nov 04, 2025 12:52 pm
Eodnhoj7 wrote: Tue Nov 04, 2025 8:53 am Please explain why you matter....please do.
That challenge sort of explains what is wrong with you. You don't realise that sane people don't consider themselves great philosophers.

You think you are in competition with the greats only because you are a delusional egomaniac and it just seems obvious to you that everything you do is really the absolute best. VA thinks along similar lines for somewhat similar reasons. It doesn't raise any alarms for people like you to that there is a dogshit little forum in the absolute armpit of the internet where every week at least five psychos claim to be the greatest ever and to have fixed the most challenging problems of philosophy. You just think those other guys are mad, but you are doing great.

AI is bad for your mental health because if you are an egomaniac already, and you ask it for belly rubs, it will give you a glowing reference and compare you to favourably to Einstein, which in your mind isn't suspicious at all. If you are inclined by way of egomania to fall for that sort of thing, it will contribute to your further decline.

There is nothing that can be done for you. This is not an intervention, I won't plead for you to get in contact with a mental health professional. This is just an explanation of what is going to go more and more wrong with your life.
You take it as a challenge...all I am doing is asking a question.

If these philosophers you claim I compete against are so great...than why should I value your opinion considering you are not one of them?
User avatar
FlashDangerpants
Posts: 8815
Joined: Mon Jan 04, 2016 11:54 pm

Re: Be careful (VA) about what you do with AI

Post by FlashDangerpants »

why did you delete the big rant?
Skepdick
Posts: 16022
Joined: Fri Jun 14, 2019 11:16 am

Re: Be careful (VA) about what you do with AI

Post by Skepdick »

The best argument for the continued use of AI is that it annoys people like FlashDangerDork.
FlashDangerpants wrote: Wed Oct 29, 2025 2:48 pm For me AI is a useful search engine tool, an annoying poison that aggressively autofills VSCode, and sometimes a quick way to get a summary to help me remember which author I am attempting to paraphrase. But VA, who once said he uses ChatGPT like a spellchecker and nothing more, has developed a concerning dependency on it as a surrogate for some role that should belong to an organic human.
There! There! He's moralizing. Shaping humans in his own image.

The fact that AI isn't a overly-critical contrarian prick is a feature not a bug - the "aggressive autocomplete" is just sentiment explication. It's precisely what you need to construct the steelman version of the tautology you are trying to express. It's constructive unlike your methods.

The wordsmith types hate it because it's automating them out of jobs. Fuck the luddites - it's the same old puritan protectionism manifesting as meatware chauvinism.

Tool usage is what defines progress. The less time we spend on dumb tasks (like fashioning the philosophical narrative) - the better off we are. Philosophical output is just plumbing in a teleological worldview.
Flannel Jesus
Posts: 4302
Joined: Mon Mar 28, 2022 7:09 pm

Re: Be careful (VA) about what you do with AI

Post by Flannel Jesus »

Eodnhoj7 wrote: Tue Nov 04, 2025 8:42 am
Flannel Jesus wrote: Wed Oct 29, 2025 12:58 pm I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c
Summary: A new study reveals that when interacting with AI tools like ChatGPT, everyone—regardless of skill level—overestimates their performance. Researchers found that the usual Dunning-Kruger Effect disappears, and instead, AI-literate users show even greater overconfidence in their abilities.

The study suggests that reliance on AI encourages “cognitive offloading,” where users trust the system’s output without reflection or double-checking. Experts say AI literacy alone isn’t enough; people need platforms that foster metacognition and critical thinking to recognize when they might be wrong.
Top comment is absolutely right:
Might this have something to do with LLMs being sycophantic (the classic "You are absolutely right!" glazing) or perhaps LLMs just being LLMs and not magic (i.e. prone to "hallucinations" and other issues which will be "fixed soon")?

I do use LLMs occasionally but only for things where I can easily verify that the LLM is correct.
VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
I have found the best way to use AI is the same manner one uses a calculator....except the calculator is a philosophical one.

Express an idea.

See if the AI can explain it back coherently with further examples and questions that come from different angles.

Ask the AI to grade in on a scale according to what metrics you choose to measure by.
None of what you described sounds anything like how one uses a calculator.
Eodnhoj7
Posts: 10708
Joined: Mon Mar 13, 2017 3:18 am

Re: Be careful (VA) about what you do with AI

Post by Eodnhoj7 »

Flannel Jesus wrote: Wed Nov 05, 2025 11:33 am
Eodnhoj7 wrote: Tue Nov 04, 2025 8:42 am
Flannel Jesus wrote: Wed Oct 29, 2025 12:58 pm I saw this on Reddit today and it made me think of VA and how he uses AI.

https://www.reddit.com/r/science/s/xzPCZvpM5c



Top comment is absolutely right:



VA seems to have been duped by the AIs over willingness to agree with him and compliment him. He took those comments to heart and made himself look a bit silly in the process.
I have found the best way to use AI is the same manner one uses a calculator....except the calculator is a philosophical one.

Express an idea.

See if the AI can explain it back coherently with further examples and questions that come from different angles.

Ask the AI to grade in on a scale according to what metrics you choose to measure by.
None of what you described sounds anything like how one uses a calculator.
Calculators calculate distinctions. Numbers are distinctions. Concepts are distinctions.
Flannel Jesus
Posts: 4302
Joined: Mon Mar 28, 2022 7:09 pm

Re: Be careful (VA) about what you do with AI

Post by Flannel Jesus »

Eodnhoj7 wrote: Wed Nov 05, 2025 7:47 pm
Calculators calculate distinctions. Numbers are distinctions. Concepts are distinctions.
Jesus Christ you're insufferable.
Post Reply