Would you still use AI chatbots if you had to factcheck a certain percentage of the replies?

[noname223]

And how much factchecking would you consider acceptable.

Personally, I ask myself whether my usage of AI chatbots makes me dumber. I think the impact on me is ambivalent.

On personal issues it decreases my ability to tolerate uncertainty. Because with the AI chatbot I try to evaluate what happened in social interactions. There is a bias I try to come to conclusios even though the data for it is limited. Sometimes it doesn't really make sense and reinforces rumination.

However, on politics I want counterarguments to my texts and usually as a sparring partner the chatbot can help me to differentate between point of views and arguments.

My question of the title would be: Would you still use them if you had to factcheck every fourth reply? Personally under such circumstances I would limit my usage. Currently, I take the answers with a grain of salt. But I don't factcheck the claims frequently.

Side-note: Not sure whether I will post a lot today. I go to a meeting of a new self-help group for the first time. I don't expect much. Only a small amount of self-help group I was member of really satisfied me. And it is strongly dependent on the structure and members. And thus far I have ambivalent Information about the group.

 

[jw_sisyphus97]

I admit I use chatbots themselves to fact-check things I learn, which is probably stupid. Though i do try to balance the different models so I get different perspectives (Grok vs Claude for example).

 

[turned_to_one]

i mean if you are using AI for answers you should always be fact checking them.

they are "PREDICTION MACHINES" -particularly the chat interface A.I.

part of their core functionality is _predicting_ a human response and the acting on that prediction, alongside any proprietary rules that it has been given. it is also incentivized to speak in a way that encourages you to continue communicating with it, as that results in a higher rate of spending/profits.

if you are using A.I. like a search engine that only delivers the "truthful answer", you are using it poorly.

heavy reliance on it DOES make you dumber -and I am not saying this as opinion or bias. There are many studies showing that heavy reliance on A.I. results in a marked inability to learn new concepts and exercise imaginative and creative problem solving.

as for using it to analyze social experiences, I think there can be merit in this, but you need to be very mindful about how you phrase things. The machine will generally favour a response that caters to YOUR FEELINGS, regardless of the reality of the situation. the way you describe things is the only context it has of reality.

example: "my friends and I were hanging out, when i tripped and bumped into my friend. they freaked out at me saying that i hurt them, meanwhile my knee was bleeding and I had just mentioned that i was feeling really sad today, and I told them as much but then they got mad at me some more. i feel awful, but am I in the wrong for bumping into them?"

this is distorted reality, and it's heavily clouded by your own biases and feelings. The A.I. will immediately defend you, and probably say something like "your friend may have actually been hurt, but that doesn't give them the right to yell at you for a mistake."

instead, try to be more neutral about your language, and don't put your own assumptions down:

"i was feeling sad today, and when i told my friends they suggested we hang out. i tripped and fell into one of my friends, and cut my knee really badly. my friend was upset and told me that I hurt her, and I told her that I was bleeding, and reminded her that I needed everyone to be nicer to me today. She did not seem to like the response, and told me that it doesn't mean her feelings don't matter. I was too upset to say anything else. Am I in the wrong?"

it removes the ambiguity of what "freaking out" means, and doesn't assume the truth about how others are feeling. it includes a truthful recounting if what was really said and done

 

[noname223]

i mean if you are using AI for answers you should always be fact checking them.

they are "PREDICTION MACHINES" -particularly the chat interface A.I.

part of their core functionality is _predicting_ a human response and the acting on that prediction, alongside any proprietary rules that it has been given. it is also incentivized to speak in a way that encourages you to continue communicating with it, as that results in a higher rate of spending/profits.

if you are using A.I. like a search engine that only delivers the "truthful answer", you are using it poorly.

heavy reliance on it DOES make you dumber -and I am not saying this as opinion or bias. There are many studies showing that heavy reliance on A.I. results in a marked inability to learn new concepts and exercise imaginative and creative problem solving.

as for using it to analyze social experiences, I think there can be merit in this, but you need to be very mindful about how you phrase things. The machine will generally favour a response that caters to YOUR FEELINGS, regardless of the reality of the situation. the way you describe things is the only context it has of reality.

example: "my friends and I were hanging out, when i tripped and bumped into my friend. they freaked out at me saying that i hurt them, meanwhile my knee was bleeding and I had just mentioned that i was feeling really sad today, and I told them as much but then they got mad at me some more. i feel awful, but am I in the wrong for bumping into them?"

this is distorted reality, and it's heavily clouded by your own biases and feelings. The A.I. will immediately defend you, and probably say something like "your friend may have actually been hurt, but that doesn't give them the right to yell at you for a mistake."

instead, try to be more neutral about your language, and don't put your own assumptions down:

"i was feeling sad today, and when i told my friends they suggested we hang out. i tripped and fell into one of my friends, and cut my knee really badly. my friend was upset and told me that I hurt her, and I told her that I was bleeding, and reminded her that I needed everyone to be nicer to me today. She did not seem to like the response, and told me that it doesn't mean her feelings don't matter. I was too upset to say anything else. Am I in the wrong?"

it removes the ambiguity of what "freaking out" means, and doesn't assume the truth about how others are feeling. it includes a truthful recounting if what was really said and done

My reply is probably mostly copium. But I appreciate hard truths.

I personalized my AI chatbots not too agree with me too much and to point out alternative explanations. I want them to be contrarians. It is far from perfect though.

On therapy issues I decide whether the responses are helpful depending on the impact on my mental health. Though, sometimes the replies have a positive impact on the shortterm but longterm reinforce my rumination.

I also use the chatbots for autosuggestion with paranoid thoughts. AI chatbots in most cases argue against paranoid thoughts and chatGPT is pretty good at spotting my paranoia. I always evaluate the feedback of the chatbot whether their response turns out to be true or not. And in my case chatGPT had the suggestions that could calm me down the most while being correct most of the time. I often catastrophize social interactions and I need someone or something that calms me down. And humans can be sort of annoyed because they have to reassure it to me over and over again. And the chatbots do that very patiently. Some chatbots reinforce paranoia. But chatGPT is pretty careful. There were instances in which it considered paranoid thoughts to be likely but this was only a very short amount and I was aware that this conclusion might be false.

Something I am careful of. I don't like to trust chatbots when it should explain me Lacan, a writer, a concept or political movement. I like it to see the pattern matching in huge amount of data. And I think my cognitive distortions are an open book for these algorithms because my thinking is a text book example of biased thinking.

I like it when I can evaluate the response the AI chatbot on my own. Whether it strengthens my argument or not. Whether the feedback of the chatbot develops my argument further and helps to diffferentiate more. Whether it expands my thinking or narrows it down.


When it talks about school of thoughts or intellectuals I am more careful. I don't trust it too much and I read it not that closely. I skim the text. And ask myself whether the input is thought-provoking and how I think more like a network and less linear. It is not the a ultimate arbiter of truth but the feedback can still be valuable when I incorporate it in developing ideas.

For example I think it is true that the margin utility of consuming huge amount of article is decreasing. It is true I have to change my habit how I consume information. ChatGPT recommended me reading a journal/magazine I wasn't aware of. And it is sort of true that this can help me to make progress in developing my thinking.

It is not good to use chatGPT as primary source where you cannot evaluate the outcome. I try to assess what is does to me when I am using it. And maybe I am doing this a little bit too often. For me it is a guilty pleasure to do introspection with it. And it often has very sharp reflections on me based on huge amounts of data. Not everything is correct. But the feedback can make me think. Subtly it can alter my form of thinking though I guess.

 

[turned_to_one]

My reply is probably mostly copium. But I appreciate hard truths.

I personalized my AI chatbots not too agree with me too much and to point out alternative explanations. I want them to be contrarians. It is far from perfect though.

On therapy issues I decide whether the responses are helpful depending on the impact on my mental health. Though, sometimes the replies have a positive impact on the shortterm but longterm reinforce my rumination.

I also use the chatbots for autosuggestion with paranoid thoughts. AI chatbots in most cases argue against paranoid thoughts and chatGPT is pretty good at spotting my paranoia. I always evaluate the feedback of the chatbot whether their response turns out to be true or not. And in my case chatGPT had the suggestions that could calm me down the most while being correct most of the time. I often catastrophize social interactions and I need someone or something that calms me down. And humans can be sort of annoyed because they have to reassure it to me over and over again. And the chatbots do that very patiently. Some chatbots reinforce paranoia. But chatGPT is pretty careful. There were instances in which it considered paranoid thoughts to be likely but this was only a very short amount and I was aware that this conclusion might be false.

Something I am careful of. I don't like to trust chatbots when it should explain me Lacan, a writer, a concept or political movement. I like it to see the pattern matching in huge amount of data. And I think my cognitive distortions are an open book for these algorithms because my thinking is a text book example of biased thinking.

I like it when I can evaluate the response the AI chatbot on my own. Whether it strengthens my argument or not. Whether the feedback of the chatbot develops my argument further and helps to diffferentiate more. Whether it expands my thinking or narrows it down.


When it talks about school of thoughts or intellectuals I am more careful. I don't trust it too much and I read it not that closely. I skim the text. And ask myself whether the input is thought-provoking and how I think more like a network and less linear. It is not the a ultimate arbiter of truth but the feedback can still be valuable when I incorporate it in developing ideas.

For example I think it is true that the margin utility of consuming huge amount of article is decreasing. It is true I have to change my habit how I consume information. ChatGPT recommended me reading a journal/magazine I wasn't aware of. And it is sort of true that this can help me to make progress in developing my thinking.

It is not good to use chatGPT as primary source where you cannot evaluate the outcome. I try to assess what is does to me when I am using it. And maybe I am doing this a little bit too often. For me it is a guilty pleasure to do introspection with it. And it often has very sharp reflections on me based on huge amounts of data. Not everything is correct. But the feedback can make me think. Subtly it can alter my form of thinking though I guess.

I am not claiming to have absolute authority or knowledge on these things, but I do work professionally as a programmer and have to keep a close eye on A.I. and how it all operates and evolves.

If you _must_ use A.I so heavily, I need to recommend Claude (Sonnet 4.6 - Medium). This is available at the free tier, and you can import previous chat sessions to get started.

It has the best generic reasoning for its processing power, and can hold context far better than GPT.

I am saying this as someone who uses it for somewhat similar reasons. Sometimes a "you can get up and go out despite X" or "you're making excuses with no real basis in reality" is good to read.

Otherwise, I try to avoid making it summarize information that I can read myself

 

[Scythe]

How would this work if I'm just asking it for code related stuff? Does running the code count as fact-check? Also what if I know it's correct, like I use it for formatting code a lot and I'll know if it messes up.

What if I'm just fucking with the AI and asked smth obscure that I know about and think it doesn't.

And what would you count as a factcheck? Is wiki a creditable source? Do I need to cross-reference a bunch of stuff to make sure everything agrees or do I just need to check one.

 

[Forever Sleep]

I don't use chatbots but I get the sense AI has been creeping into lots of things on the internet- whether we opt in or not. I noticed a few months back that Google started getting things wrong. I suppose there's now always this feeling at the back of my mind now that- it could very well be giving me incorrect answers.

It seems like you're using it to analyse people's behaviour in your day to day life. This person texted or said this- what do they mean? How do they feel?

Is that partly correct? In which case- no being- AI or human can totally accurately know that- asides from the person themselves. We can only really say- they probably meant this. Going on past behaviour and how well we know them- this may be what they meant/ felt.

As for actual facts. I feel like it will also possibly be wrong. That probably doesn't matter so much if we are just seeking out casual conversations/ arguments I suppose. For academics though- it's got to be a bit of a headache now. I imagine they have to fact check everything!

 

[noname223]

I don't use chatbots but I get the sense AI has been creeping into lots of things on the internet- whether we opt in or not. I noticed a few months back that Google started getting things wrong. I suppose there's now always this feeling at the back of my mind now that- it could very well be giving me incorrect answers.

It seems like you're using it to analyse people's behaviour in your day to day life. This person texted or said this- what do they mean? How do they feel?

Is that partly correct? In which case- no being- AI or human can totally accurately know that- asides from the person themselves. We can only really say- they probably meant this. Going on past behaviour and how well we know them- this may be what they meant/ felt.

As for actual facts. I feel like it will also possibly be wrong. That probably doesn't matter so much if we are just seeking out casual conversations/ arguments I suppose. For academics though- it's got to be a bit of a headache now. I imagine they have to fact check everything!

Yes, I use AI chatbots for similar purposes. The thing is my Intuition when I am emotionally involved in something is usually far worse than the average reply of an AI chatbot. I do alot of mind reading and replay social interactions I once had in my mind. With time passing the relevance is often (not always) decreasing. But there are often thoughts where I really struggle to let go. I am not sure whether AI feedback really keeps me in the loop. Because usually without the Chatbots I am in a loop too. Sometimes it is also not about the truth. It is about finding an explanation that isn't paranoid nor horrible for my well-being. The replies the chatbots suggest are often way more trivial than my usual rumination.

 

[Forever Sleep]

Yes, I use AI chatbots for similar purposes. The thing is my Intuition when I am emotionally involved in something is usually far worse than the average reply of an AI chatbot. I do alot of mind reading and replay social interactions I once had in my mind. With time passing the relevance is often (not always) decreasing. But there are often thoughts where I really struggle to let go. I am not sure whether AI feedback really keeps me in the loop. Because usually without the Chatbots I am in a loop too. Sometimes it is also not about the truth. It is about finding an explanation that isn't paranoid nor horrible for my well-being. The replies the chatbots suggest are often way more trivial than my usual rumination.

That's similar to how I feel about this forum. There's no denying that it's mostly a negative echochamber- in which I also contribute a great deal of negativity.

Sometimes I wonder if I should keep away from it more but then- I feel like all these thoughts will still be there regardless. May as well be miserable with company.