What are the bad sides to chat gpt ?

[AceKitten]

Is it to do with too much water ? Personal information being taken ?
or anything else ?

Huge energy use for processing, particularly for AI-derived images and vids. Lots of water needed for cooling. Lots of other issues such as bias, hallucinations etc. I love ChatGPT and use it a lot but there are definite issues

Security is an issue.
AI is essentially continuously learning and takes information given to it, to also learn from with the basic settings. As such you should never add personal data when asking questions.

The environmental impact is also an issue, though many if the really big companies like Microsoft are working on their environmental policies and many are separating out AI so they have the overall environmental reports and how AI itself is impacting.

Bias and misinformation is also an issue. Anyone can write anything on the internet, as the AI system searches it can easily come up with misinformation.
In addition based on how it's programmed it can also have a bias to the answers.

Finally it's reported that it is worse when it comes to maths and legal answers.

AI can be fantastic, but if you know the pitfalls you can double check information when needed - but really you should do this with any information you're given. It takes me back to history lessons as secondary school, who wrote it, when and why.
Ask yourself these questions and you can go too wrong and don't use it for something a basic search engine can do

Yet another example of MN educating me.

I hadn't a clue what " too much water" had to do with Chatgpt. So I googled it and it was quite a revelation to me.

I've heard of a lot of negative aspects to Ai but not this. So thanks for raising my consciousness and teaching me yet again to take more notice of things which really impact our lives.

I find my husband uses it so much and it’s like it’s his go to…before even google

'Hallucinating' information (i.e. giving made-up info)
Has sycophantic tendencies (tends to agree with whatever you say, which can be harmful)

Don't use it for important things like finding health information - often takes info that may not be relevant (e.g. from other countries), extracts random bits of info from places that are not designed as patient information etc

It always needs human oversight - information always needs checked.

I occasionally use it (not chat gpt, my browser's own AI assistant) instead of googling, but I'm wary of misinformation. I asked it recently to summarise a book I wanted to read, but was so wordy I was struggling to get through it. It gave me a neat precis of each chapter which bore absolutely no resemblance to the actual book! I'm careful what I ask now and fact check everything.

Oh where to start!

It’s already inaccurate and as time goes on what it learns from will be other inaccurate AI and eventually no one will be easily able to decipher what’s factual and what isn’t.

It doesn’t ‘create’. It’s just a giant web crawler so all it’s doing is collating, not thinking or making new. And because it’s collating from everywhere it includes all the racist, sexist, ableist crap in with the decent content without knowing that it’s wrong or inaccurate.

It has its uses but it’s not ‘intelligence’, it’s a summarising machine.

Big confirmation bias too.

I find it concerning the people who use it like a friend, and talk to it. As much as I understand being lonely, and that it may fill a gap, it isn’t real, it’s just a computer program, but I think more and more people are starting to treat it like a real person.

Have you been reading this thread, @AceKitten? 😁

https://www.mumsnet.com/talk/_chat/5450430-i-have-been-arguing-with-chat-gpt-and-wondering?reply=148780265

Everything other posters have said, and for me, it's reducing people's critical thinking. It presents nonsense or biased information, and people take it as gospel. I use -ai on all my Google searches so at least you don't get the ai summary as the first result.

That's not a plus for ChatGPT, but just a reflection that Google has got more shit in the past decade. It may even be in a death spiral now it's punting "Gemini" (wrong 90% of the time) as the answer.

E2A: for now you can add "-ai" to the end of your search, or use https://www.google.com/?udm=14 to avoid "AI" summaries.

sycophantic Tendencies so accurate !!

No but I will do !!

Yes i totally agree with that
esp when it become your go to

sad when my dh had a hospital procure the other day and instead of coming straight out and explaining it
fresh
was waiting in waiting room

he was taking a photo of the later and report uploading it to his chat gpt as he has an account and all that so all his info is stored even more

and he was looking at chat gpt before he even spoke to me to tell me, his wife, mother of his dependant children what the dr said

it’s just bamn straight into chat gpt

I’m a huge fan of AI and am using it multiple times a day. I fact check sources and wouldn’t blindly trust anything in the same way I wouldn’t blindly trust something I’d googled. It can often help critical thinking by offering different and opposing views if you ask it to. It helps me work develop my thoughts and opinions. As long as you ask the right questions and are always mindful of the fact it’s ai then I think it’s really useful.

My Mum is 87 and lives alone. Her eyesight isn’t great so sometimes she finds it difficult to work out how to do things. AI is so useful for her.

It doesn't always give the full story. Nine times out of ten when I ask it for some information then query something its provided I get a "great catch BasilParsley, you're quite right..." and will then adjust the answer accordingly.

It has been built on wholesale theft of intellectual property. Ripping off other people’s hard work to train the large language model.

I also think it'll create an incredibly lazy generation and I think in general, we will lose some great skills across the board.

Also, the sci-fi fan in me thinks it's a recipe for disaster..! 😂

But in all seriousness, I think it does way more harm than good.

There aren't any really.

Just as long as you apply your "healthy scepticism" filter.

I've used it for stuff I know something about and while it is often not perfect the summary it gives is often close enough for someone who knows nothing about the subject.

I do some coding and it can be great for teaching you how to do stuff, producing code snippets that you can modify.

It can give you a quick overview on something and often knowing something is better than knowing nothing at all. I wouldn't use it to teach me how to pilot a jet plane or do open heart surgery though.

My gut feeling is that its main detractors are people that are scared of it. Mostly scared for their jobs. And quite rightly so.

I dislike getting ai as the top google result as it’s often wrong can someone please explain how to stop that happening thanks

Someone upthread said to include ‘-ai’ in your search terms. I’m going to try that…

Ah so you type it in each time then ?

This. People take it as fact but it isn't. Frequently it is simply wrong.

Lazy journalists or lazy students (or anyone else not doing sufficient research) is going to find themselves looking very foolish.

Anything you feed it becomes information for future users. DH just sacked someone for putting in sensitive business info as part of putting a tender bid together. The fact that ChatGPT had this info was raised to them by a competitor! 🤦🏼‍♀️