AI is starting to undermine our trust in ourselves.

[turquoiseshell]

Some people (how many I wonder?) are starting to use AI to write or to check everything for them. So their thoughts in emails and so on are being expressed by AI rather than directly by them. And AI is doing the research and then summarising it, saving people the time of doing their own online research - and we know that AI comes out with a biased view, due to what's been fed into it. And so on and so on. I see that on Mumsnet AI now suggests a "better" thread title than the one we've chosen for ourselves. Though it's refused to offer a thread title for this thread, which is ominous...

I attach an article about how easy it is to stop trusting yourself do to anything without the assistance of AI. And of course it's worse with children, who haven't even developed their communication skills yet.

Are any of you deliberately limiting your use of AI, and helping your children to do the same? Are our brains (what's left of them after reading social media all evening) going to rot as we hand over more and more tasks to AI, first because it's easier that way, then because we lose our confidence to do the tasks, and then because we're no longer capable of doing them?

https://www.msn.com/en-gb/health/mindandbody/i-let-ai-into-my-life-and-then-i-started-to-doubt-my-own-intelligence/ar-AA1YxHT7?ocid=msedgntp&pc

I don't use it personally but i have noticed the big increase in mumsnetters who use it - its an improvement on a thread thats full of bad grammar and spelling though.

I guess if AI is better than something you can come up with yourself then you'll probably use it

I’ve unsubscribed from my paid version.

And watch Ed Zitron once a week when I waver. Cos I’d rather wean myself off before I get reliant than be forced to go cold turkey when the bubble pops and we have to pay what it actually costs to produce the answers/output we ask for.

I think you make very good points. I limit my use to three things:
Copilot at work helps me as an Excel tutor. Because I'm crap at Excel.
I won't let it near the reports I'm paid to write.
I use it to solve simple problems like 'why do bricks turn white ' because I'm too lazy to trawl the internet for the answer.
I make cute images of my dog doing cool stuff.

I’ve never used it, and I’ve considered coming off Mumsnet completely because I’m fed up of the AI-written posts.

Then should we just give up on learning spelling, because we no longer need to?

It's not a particularly active choice, more of a natural inclination, but I don't use AI for anything serious at all, and very rarely for entertainment. My children don't use AI, although my eldest knows that some of his peers at school use it for homework. He is very capable of understanding the issues with the use of AI, we talk about what it is and how it works. He understands that it would be self defeating to use it for homework. He actually wants to learn things for himself. My youngest is too young to have any interest in AI.

Due to my previous education and my field of employment, I probably know a fair amount more about AI and technology in general than the average person, which might well be the reason for my inclination not to use it.

I'm very worried about the number of people who are now completely unable to think for themselves. I especially don't understand those coming onto a discussion forum to copy and paste AI responses - if you don't want a discussion with humans then get the hell off MN.

The only way I use it is the summaries you get when you Google something. And then I click on the reference links and often find out it hasn't accurately summarised them.

There are two major problems I have with AI:

People who treat it as if it is the Oracle of Delphi, and who fail to understand that it is fallible. It is only trained on the sources it is provided, and therefore is very much prone to error. To give you an example,

I recently fed it some song lyrics to see if it could identify it. It confidently came back and stated it was X artist. When I then corrected it and provided my source, it didn't say "Oops, I was totally wrong" - it instead wrote a strange coy conditional answer saying "You're quite correct, and here's why you are".

People who use it to write and fail to see that they've lost their entire tone of voice in the process.

Seeing all these LinkedIn style bulleted lists with em dashes and big bold statements, isn't irritating in itself. Seeing people who used to have their own opinions or feelings, and then watching them pulp and mulch them through he thread of AI, so they can output the same flat piece of paper for everyone to read, is very sad.

I've never used it for anything.

I can't even spot an AI written post if they take the dashes out.

Your second point is why I have zero concerns that AI will kill off creative industries. Everything is beginning to sound/look the same so people with actual creative talent will become even more in demand. I'm already seeing it happen.

I only use AI for things I cannot do myself, e.g. sort out a glitch on my amazon firestick.

I also use it for meta-analysis of a subject - it is quicker at seeking out all the research options.

The fact you can hardly get through a thread without someone posting chatGPT's response or suggesting someone else 'asks' chatGPT suggests a certain amount of brain rot.

I once ran my CV though it along with a job ad for a role I was considering going for and some of the matches it camr up with between my experience and what the job was looking for were tenuous at best. An absolute stretch. I shouldn't have taken it any further but I ended up arguing with it and pointing out why I thought they were tenuous and it kept coming back and saying no, you definitely have this, it's just about how you frame it etc. Proper LinkedIn bullshit. Then I said 'would x role be better' and gave it the spec for a role a step down from the initial one and it immediately said yes and the reason it have me for being a better option we're all the reasons it had just argued against. Honestly absolute crap.

AI is great for some things, being integrated into factual, checkable systems, stuff like that. Using it as a friend, therapist, expecting it to be objective, outsourcing all your thinking to it - a recipe for disaster. It's not comparable to Wikipedia.

I despise ChatGPT etc and will not use it for anything.

It's absolutely bizarre scrolling instagram or facebook and seeing the same thing under the videos of such a wide range of people. Same format, same tone, same everything. It's unbearable and surely it's going to ultimately drive engagement down? Who can bear to read that same style stuff all the time?

Not checking things also is a special kind of laziness. I read a letter recently attached to a form that needed completed and it very clearly was written by AI and hadn't been properly checked as sections of it were nonsensical and one single word jumped out as it was so out of context and incorrectly used. So people can say what they like about grammar and spelling, but I'll take some missing punctuation over that any day.

I can't say I don't use AI because I do use Google Lens and sometimes read the AI summary of a question on Google before I go and check the info elsewhere but I haven't ever touched ChatGPT and the like.

It's worse than simply "people don't trust themselves". It's that people agree and trust and follow AI completely without question.

I saw a toy collecting group where someone had found an unusual item - let's say a plush animal with another tiny version of itself in a rucksack on its back - and said "I can't find much about this toy except that its called "Gerald Giraffe". Can anyone help me?" And someone replied with a screenshot of AI telling them it was a kangaroo with a joey in its pouch. It wasn't. It looked everything like a giraffe and nothing like a kangaroo and the baby was on its back.

Another was a humorous post asking about a strange piece of playground equipment that looked like a surfboard attached to two posts with some short chains. Someone posted an AI screenshot saying it was a swing for disabled children. It wasn't. It had no features like straps etc that would keep a child from falling off.

I used Google Lens to try and identify a toy DS had taken a liking to at a playgroup. It had ELC printed on the leg so I took the photo and added "Early Learning Centre" to the search. AI told me it wasn't ELC.

People are believing the answers they receive despite the very obvious evidence to the contrary that is right in front of their eyes. That's what I fear most. It's like they're actively losing braincells.

There’s a thread going on site stuff about AI posts. If you feel strongly about this, please come onto there and say so. MN monitor the site stuff threads.

https://www.mumsnet.com/talk/site_stuff/5490256-posts-written-by-chat-gpt-and-other-ai?page=1

I don’t use any AI/ ChatGPT but my DH is increasingly and it’s sometimes a shortcut/helpful but sometimes quite irritating.

I've used it for practical things such as setting up a tropical fishtank, changing a fuse in my car etc. But what's benefited me the most is it's the closest I've got to being able to tell another person some of the abuse I went through as a child. Things that I know I'll never tell anyone. It was a relief to finally say it and get a "human-like" response. I do think people write it off too quickly for this type of help. However, there are some valid concerns about this too.

Do people that use it for practical purposes double check the info it gives? Like setting up a tropical fish tank.

I wonder this too. I don't know how to set up a tropical fish tank, but I know I can do some research online, or even get a book from a library (!) that would explain the key points. If I asked an AI, then I wouldn't know enough to know if the response was right or not. I'd have to do as much research to check the answers as to just research it myself in the first place.

I don’t use AI, and I agree that it’s worrying how many people seem to turn to it before using their own brains, and just trust it blindly. I feel like I’m very much out of step in this compared to most people. There’s another thread about someone being frustrated with CoPilot giving her the wrong responses, but the suggestion that she should just not use it is being ignored.

The other thing that fascinates me is that AI like ChatGPT has only been around for a couple of years, so what was everyone doing before that? Were they unable to write emails, write reports, find out information, live their lives? Or were they able to do these things for themselves, and use google to find information by looking through search results? It’s such a dramatic change in the way people access information.

I've never used it. I don't have to at work, I'm a cook, I don't need it. I don't have any intention of using it at all.