Weststigers Wasteland

[Nutz]

It's midnight and have had my old hip replaced by a new spanker, I start footy training in the morning

Like most AI programs, I assume ChatGPT relies on human input for facts which surely opens it up to inaccurate information at times. Who fact checks ChatGPT as no program is foolproof.
I like it when people quote information directly from the horses mouth. BUT how many times do you read....I didn’t say that.
For a quick heads up on matters that I have no idea on, I use Wiki. It's quick and generally helps me get a rough idea.
A lot of debating gurus like yourself and TooheysNew probably detest Wiki but I find it quick and simple. I use it to start the ball rolling. It often has links to the horse.
The real issue I feel is for each generation to educate people like me on how to ensure our planet is habitable and enjoyable until we are all consumed and return to dust.

 

[TooheysNew]

On the contrary, I think Wikipedia is a great resource, provided that it's used effectively. i.e. don't just read the article. Dig through the sources and read those instead. Actively look for information that disproves your theories, and attempt to validate those.

Using ChatGPT for this is not the correct approach though, as it has no understanding of the content or the context. It is literally just using probability tools to work out what word (token) should come next.

 

[Nutz]

Just for some context I will add how coral reefs are handling climate change:-

• Coral reefs are declining worldwide in both coral cover and biodiversity.
• Even reefs that appear “healthy” may have undergone species loss, reduced complexity, and lower resilience.

This seems feasible. Creatures and plant life have continually adapted and evolved since Adam wore nappies.
To say that an increase in coral activity is due to lesser complexed coral taking over, or surviving has merit.
Coral needs Saltwater and that's why the mainstream coral doesn't exist where fresh water drains into ocean.
The poor quality coral can exist in non ideal conditions. I don't know the name but I've seen it. That coral offers little or no benefit to the reefs eco sys. Eg coral eating fish won't touch it and shite the contents out and doesn't filter the water.
Fact check. I've seen this and have talked to locals about it.
Coral generally needs clean water to thrive. Fertilisers and waste water can be seen emptying into the ocean for several km which is void of coral.
Fact check. I've seen it patrolling in helicopters and talked to CSIRO marine biologists about it, that was in 2000

Usually, coral will not grow in deep water as it needs sunlight.
However the depth of water for healthy coral deepens if the water is ultra clean so the rays are not cloudy.
Fact check. I've seen this scuba diving with an instructor and have talked to locals.
Just some points about the horse's mouth comment I made earlier. If it quacks like a duck. No quotes or links.

 

[Tiger05]

On the contrary, I think Wikipedia is a great resource, provided that it's used effectively. i.e. don't just read the article. Dig through the sources and read those instead. Actively look for information that disproves your theories, and attempt to validate those.

Using ChatGPT for this is not the correct approach though, as it has no understanding of the content or the context. It is literally just using probability tools to work out what word (token) should come next.

You aren't being fair though. Those guys posted misinformation and it was clearly stupid once you figured it out but you have to figure it out. I couldn't tell just from a simple Twitter tweet but ChatGPT figured out they had used the hottest day at one point and the coldest day at another point.

Could you have figured that out that quickly ? I bet you couldn't.

This brings up something that I think we will both find interesting and that is critical thinking skills. One thing I've noticed in nutrition and in climate science (I made this mistake) is thinking we know how something works mechanistically but the results don't conform to the mechanistic idea. That means we don't understand the mechanism as good as we think we do.

Your comments on this thread show a lack of understanding of how ChatGPT works. Can you explain how ChatGPT picked up on cherry picked data so quickly and easily and it got it right. It's gotten heaps right that I've fact checked as well.

 

[TooheysNew]

You aren't being fair though. Those guys posted misinformation and it was clearly stupid once you figured it out but you have to figure it out. I couldn't tell just from a simple Twitter tweet but ChatGPT figured out they had used the hottest day at one point and the coldest day at another point.

Could you have figured that out that quickly ? I bet you couldn't.

This brings up something that I think we will both find interesting and that is critical thinking skills. One thing I've noticed in nutrition and in climate science (I made this mistake) is thinking we know how something works mechanistically but the results don't conform to the mechanistic idea. That means we don't understand the mechanism as good as we think we do.

Your comments on this thread show a lack of understanding of how ChatGPT works. Can you explain how ChatGPT picked up on cherry picked data so quickly and easily and it got it right. It's gotten heaps right that I've fact checked as well.

I'm being totally fair. I'm simply informing you of how misguided your approach is.

Could I have come up with potentially incorrect information that quickly? No. But why would I want to? If you want to be right, be right. If you want to be fast, be fast. You've got no guarantee of both if you use ChatGPT as your sole fact checking source.

The part where you say I don't understand how ChatGPT works is laughable. I build large language models as part of my job. Of course I know how they work.

 

[Tiger05]

I'm being totally fair. I'm simply informing you of how misguided your approach is.

It works and it works well.

Could I have come up with potentially incorrect information that quickly? No. But why would I want to? If you want to be right, be right. If you want to be fast, be fast. You've got no guarantee of both if you use ChatGPT as your sole fact checking source.

You argument falls apart here. The reality is ChatGPT knew what was going on and it did it quickly. It was in this use case absolutely awesome.

It got it right as well.

So it somehow reached the right answer quickly. You can't knock those results.

The part where you say I don't understand how ChatGPT works is laughable. I build large language models as part of my job. Of course I know how they work.

My take is you are suffering from hubris and the Dunning-Kruger effect and it's cool.

I love telling this story. I studied economics at university and some environmental economics subjects. I studied climate change. At that point I believed climate change was fake. It makes sense mechanistically and I've seen lots of physicists say the same thing. Carbon captures waves in certain bands. The band that carbon captures is really full. It's in the high 90% range. Therefore climate change is a fraud.

Fast forward 30 years and I still didn't understand how this works. Then I saw a YT video by Sabine something or other and she explained this issue and she stated she had the same incorrect understanding as I did.

The reality is the band slowly expands. It's not as simple as my understanding of the issue was. That slight increasing of the capture of the wavelength is what is causing climate change.

I'll add another little story. I worked in IT for a big Bank. I worked on the implementation of Hadoop into the bank and I worked with top tier experts in the field. I also studied some of this on coursera. I don't think I'm an expert though. I just use the tools and if they work they work.

I'm not uneducated on IT. If you could explain the issue to me and explain why I am getting the right results then you might have a theoretical point. You haven't gone anywhere near to explaining this. I'd love to hear as well. I don't think your knowledge is up to par on this topic. I don't think you actually understand it like you think you do. That is the dunning kruger effect in action.

 

[TooheysNew]

That's a whole lot of words to say that your critical thinking skills are terrible, and that you can't adapt your viewpoint in light of new information.

 

[TooheysNew]

Here, I'll make it very simple for you. Here is a help article from the developers of Chat GPT, stating that it is not always factual. I'll even highlight the key takeaway for you to make it super easy.

---
ChatGPT is designed to provide useful responses based on patterns in data it was trained on. But like any language model, it can produce incorrect or misleading outputs. Sometimes, it might sound confident—even when it’s wrong.

This phenomenon is often referred to as a hallucination: when the model produces responses that are not factually accurate, such as:

* Incorrect definitions, dates, or facts
* Fabricated quotes, studies, citations or references to non-existent sources
* Overconfident answers to ambiguous or complex questions

That’s why we encourage users to approach ChatGPT critically and verify important information from reliable sources.

...

Practical tips for using ChatGPT responsibly
Use ChatGPT as a first draft, not a final source. Please check important information.

Always verify quotes, data, technical information or references to external documents.

Use available tools like search or deep research and check sources when accuracy matters by visiting links directly.

Encourage critical thinking—especially in educational settings.
---

https://help.openai.com/en/articles/8313428-does-chatgpt-tell-the-truth

This is saying exactly the same thing I am. The responses are based on tokens, they are an approximation of intelligence, not real intelligence, and you should use other sources to critically verify the information that it gives you.

Using ChatGPT for your fact checking and standing by it as a clear source of truth shows who here really is subject to the Dunning-Kruger effect. Sorry if that is upsetting.

 

[Tiger05]

Using ChatGPT for your fact checking and standing by it as a clear source of truth shows who here really is subject to the Dunning-Kruger effect. Sorry if that is upsetting.

It's not upsetting at all. I am completely calm and relaxed because I don't have a big ego and I am trying to learn.

Your argument is an example of what is called the strawman fallacy.

I'll just add in relation to the rest of your post I really don't see the issue at all. I'm not sure but maybe you are confusing my use cases with different use cases.

I am going to repeat what I have stated again and let's see if you can possibly make a point that is relevant to what is being discussed. I want to learn, I am willing to learn, I would be honored if you could help me out.

I have used ChatGPT on multiple occasions as a fact checker and it has worked pretty well.

Here are some of the use cases I have used it for:-

1. On this thread there were some classic examples of misinformation being spread regarding climate change. ChatGPT quickly and efficiently figured out what was happening and point it out. I asked if you could do this more quickly and you responded no. This in my opinion was amazing.
2. I have used ChatGPT to check and explain several nutritional papers and ideas on multiple occasions. The results have been exceptionally good. There was one issue but I think it makes sense and I'll explain the nuance. It requires some detail. I queried the differences between animal and plant based protein. Protein is basically a molecule of 20 different Amino acids. We have to eat 9 of these Amino acids. Animal protein has these 9 amino acids in the right combination to create a protein molecule whereas plant based proteins don't. The nuance is that if you eat a wide variety of plant based foods it doesn't matter (especially beans). ChatGPT stated the protein in animal products was more complete but the details matter because more animal products leads to significantly lower chances of making it to a healthy 70 yo. I queried ChatGPT on this though and they relayed the nuance to me and admitted they didn't provide the best answer.

--> Based on these tests I think ChatGPT works great as a fact checker (you can use another word if you want too) for my use cases.

Just to be extremely clear I am not stating anything else about ChatGPT. I am stating it has worked exceptionally well for my use cases. It has been exceptional in that it has been able to check stuff so quickly. I think it would have been exceptionally hard for a human to clearly articulate the misinformation in those climate change posts. I couldn't do it that quickly and I believe you stated you couldn't as well. I will be careful using ChatGPT as a fact checker but I have yet to have a use case that has been a signficant issue. Maybe a better way to phrase this is do you think the answers that were provided to me were wrong ? If they were right why were they right ?

--> If I haven't been clear in anything or you need any further clarifications on the points I am making please ask. No problems at all.

 

[TooheysNew]

Re-read my original post. I make no comment on any of the stuff you're spouting, nor your viewpoint.

My comment was that using ChatGPT as a fact checker is batshit insane. It is not that. It is a language model that generates words based on tokens and information it has been fed, both fact based and otherwise. For that, it is great, provided you have an acceptable margin for error - because it makes plenty.

You seem hung up on proving how smart you are because you use it as a research tool. You're completely missing the point. It might get things right, maybe even often. But the fact remains that it is not a reputable tool for facts, because it makes a lot of mistakes.

If you're happy to be potentially wrong, keep using it as your fact checker. Just don't try to tell us how much better it makes your information, because it does the opposite.

 

[Tiger05]

Just don't try to tell us how much better it makes your information, because it does the opposite.

I want to be 100% clear I am not stating my information is better because I'm using ChatGPT to fact check those posts. I am simply stating I am using ChatGPT to fact check that information. The issue is the quality of information produced and not where it is from.

I am stating that the misinformation regarding climate change that was posted on here is misinformation and the information I posted is not misinformation. If you believe this to be incorrect can you please detail where it is incorrect. I am 100% confident in what I posted.

 

[TooheysNew]

OK, we might just leave it there, mate.

You've just said you are using ChatGPT to fact check information, which is backwards, and flat out the wrong approach when it comes to critically analysing information. No credible person will ever suggest this approach. OpenAI themselves don't recommend this approach - in fact they clearly recommend the opposite, as I've mentioned above. I'm not sure why you're ignoring that fact.

You can be as confident as you like in what you've posted. That's beside the point. The point is, that telling people to fact check using ChatGPT is misleading at best, and dangerous at worst.

 

[Tiger05]

OK, we might just leave it there, mate.

That makes sense. Let's just agree that I provided accurate information in a quick and timely fashion.

 

[TooheysNew]

Absolutely. We will agree that you suffer delusions, lack critical thinking skills, and have poor comprehension skills to boot. Best of luck with your misuse of AI.

Have a nice night.

 

[Tiger05]

Absolutely. We will agree that you suffer delusions, lack critical thinking skills, and have poor comprehension skills to boot. Best of luck with your misuse of AI.

Have a nice night.

Thanks - coming from you that is real positive praise.

If you have the ability to prove that any of the information I provided wasn't accurate can you please do so. If you can't well then the assumption is that the information was right. That information conforms to my understanding and I am educated on that topic so we are cooking with gas.

You have a good night too. Sleep well.

 

[TooheysNew]

Thanks - coming from you that is real positive praise.

If you have the ability to prove that any of the information I provided wasn't accurate can you please do so. If you can't well then the assumption is that the information was right. That information conforms to my understanding and I am educated on that topic so we are cooking with gas.

You have a good night too. Sleep well.

And once again you've missed the point. The accuracy of your one post is not the question. You're trying to gaslight and move goalposts, but you're not smart enough to pull either of those off.

It seriously amuses me how far you are from comprehending a very simple point.

If you're not trolling, you're very dumb.

 

[Tiger05]

And once again you've missed the point. The accuracy of your one post is not the question. You're trying to gaslight and move goalposts, but you're not smart enough to pull either of those off.

It seriously amuses me how far you are from comprehending a very simple point.

If you're not trolling, you're very dumb.

I'm just trying to be clear because I think I already responded to the same ad-hominen bullshit previously but we'll do it again.

I bolded and italicized it this time. Does that make it easier to understand ?

If you have the ability to prove that any of the information I provided wasn't accurate can you please do so.

That is the only question at hand. I think we both know the answer.

Is the information accurate ? Yes. End of story.

 

[TooheysNew]

You can bold that until you're blue in the face. That still doesn't make it the point.

Re-read my initial posts in here. Or have chat gpt read it to you, if you struggle with your own original thoughts. Good luck.

 

[Tiger05]

You can bold that until you're blue in the face. That still doesn't make it the point.

Re-read my initial posts in here. Or have chat gpt read it to you, if you struggle with your own original thoughts. Good luck.

Just to be clear you are stating you don't have the ability to prove the information I provided was wrong.

Just to re-quote it again. Maybe the third time is a charm.

If you have the ability to prove that any of the information I provided wasn't accurate can you please do so.

I'm going out on a limb here - I think you are full of shit. I think you spoke like a big man but you can't back it up. This is multiple times I've called you out on your bullshit and you haven't backed it up once.

You mentioned trolling previously. You are trolling. You talk a big game but you are full of shit. Your opinion is bullshit.

 

[TooheysNew]

Think what you like. Your opinion means little to me. You're a few pixels on a screen.

It's not that I don't have the ability to point out errors in your posts. That's not relevant. Not once have I said that there were errors in your shit above. To be honest, I haven't read any of them because it's regurgitated bullshit that I have no interest in. I've not engaged in any of them.

You're just trying to gaslight me into thinking that was the original issue again because you've been caught looking silly because you genuinely thought LLMs are a legitimate source.

Laughable.