Aibu - using AI?

[finsberry]

Im using AI for a lot of things and have become very dependent on it. Here are some examples

• putting ingredients in and asking for recipes
• translation
• filtering potentially triggering news stories like the r* academy that I don’t want to google
• Frequently for work
• for general quick questions
• occasionally to rewrite emails/ messages when the situation is sensitive

Aibu to think everyone is using it to this extent? I am using multiple times daily and pay.
yabu - not using it
yanbu - using it

I don't use it for things I can do myself i.e. for googling questions, planning itineraries etc because I'm aware of the damaging effects to both my critical faculties and the environment. I do use it to do things like check code but I never rely on it completely because it does hallucinate.

I also don't use the big US models or anything that's retaining my data for nefarious purposes https://quitgpt.org/

Interesting that there's a big thread of people wondering about the impact of retirement on cognitive abilities, and other threads where people are apparently unconcerned with the impact of outsourcing the most basic of tasks on cognitive abilities.

I know which I think is more damaging.

If you ask it about a subject you are familiar with you will catch it hallucinating sometimes immediately, sometimes after a few minutes.

So imagine that it is doing that about every subject, it's just if you don't have enough knowledge about it you can't catch it out.

Do you not think that fear and defensiveness is a normal and rational response to something that is being flaunted as a replacement for human intelligence and purpose, and that is likely to destabilise every aspect of our lives faster than we can mitigate for on practical levels, and has the ability to exacerbate mental instability which is a growing by product of modern life? And that apparently we will have no choice but to accept if we want to survive?

No, not me, still using my brain...

The newest models aren’t hallucinating anymore. The rate at which they’re improving is exponential. You can choose to use a newer model but the default will be the older ones which are quicker.

Yes, I've asked it about things to do with two areas of specialist knowledge I have. It comes out with absolute shite. Some of the things it suggests are downright dangerous.
I also asked it to come up with a list of books relating to long-distance cross-country skiing, when I naively thought it might be able to come up with suggestions for books I haven't read yet. It came up with a list of 10 books, 3 of them were real books which I had already read and the rest of them were just made-up titles with made-up authors.

I think that there may be useful applications for it as it develops but I also think it end up contributing to cognitive decline if people are just using it for things they would be perfectly capable of doing themselves.
I tutor children in Maths and English. They are all using ChatGPT for everything. Concentration is getting worse and they have become so reliant on it to do homework that they are getting very bad grades in exams when they aren't able to use it. It's something that needs to be looked at urgently in schools (and universities) because it's completely changing the way students are behaving and how they learn.

I find it useful, in my workplace, it will find coding bugs in seconds I would otherwise struggle with for days. I have asked it about recipe variants, astrophysics, movie realism, UK geology, I have constructed a solid resistance training routine and nailed down my technique on the compound lifts with its aid and I am making much better progress than I ever did in the past. It has been good for relationship issues, low self esteem and psychologically dealing with tossers in society who persist in inflicting low level friction on me on a daily basis. It is a tool and a useful one if occasionally flaky but the benefits for me far outweigh the flaws. It can come up with answers to questions far far quicker than I could Google them, and Google is as likely to come back with crappy nonsense as AI. Using it is not a alternative to thinking as some narrow minded people would like to believe and in fact, that is where AI shines, it isn't influenced by stupid irrational emotions.

I use it every day.

For now AI, won’t take our jobs, but they will be taken by people who know how to use AI.

It’s taken over the repetitive, time-consuming work and is allowing me to focus on the areas where I can add value.

I'm using my brain too but I am using it to create Agents.

If your using it regularly at work won’t be long till your employer clicks on and decides to replace your job with it. Stop using it.

I use it occasionally at work to save time but try to limit myself to that. Also found it handy for designing spaces in our new house recently so have ideas on what to tell carpenters etc. Tried asking experts to come and give us options but they expect us to know what's possible and we don't, so this has been a handy tool there!

Others I know don't seem to be able to make a decision or plan without checking with AI first, which feels a bit alarming. Environmental impacts also a concern.

How can I trust if it or isn't?!

It has been brilliant for spreadsheet work (formulas and data analysis).
With very careful choice of what tool you use I'd say it can be helpful for going deeper into (e.g.) blood test results if faced with a page of them where things are technically normal, though presented using a lot of different scales & measurements.
You could summarise meeting papers with (almost) no use of the letter 'e' if you were so inclined.
On one occasion I have had it not be able to determine what time the first train to terminus in London from X commuter belt station would be, that could conform to the off peak rules rather than the peak, and I think this is quite funny. But it learns all the time.
I'm uncomfortable with the 'if you are not paying to use it, you're its food chain' sense, though.

A few months ago it felt like my manager was saying 'run it through copilot' or 'you can ask copilot' on a weekly basis. Since I would have to check every line of output, what would be the point?

A database with intuitive AI integrated into it is very different from that kind of thing. Just chucking every thought into copilot is garbage.

The work one is potentially unreasonable, depending on what you're actually doing.

Not worried about AL nor do I fear it. I want to keep my brain sharp by using it and not relying on someone / something to do my thinking for me. I'm not a child who needs to be told how to think.

I use it for information around tech where I can't really find an answer and it's been very useful as it brings everything on the net together with useful information and pointers. I don't use it for things like recipes, hotels, etc.

If it saves or money I’m all for it. I’ll like to use it more for simple tasks…

If you’re getting crap out of AI you need to work on your prompts.

LLMs don’t understand your intent so you have to give specific instructions. If they are vague, incomplete, or poorly structured, the output will reflect that. It’s something worth practising.

I use it for my Etsy shop - it does make product descriptions a bit ‘samey’ but I’m not too bothered about that. I also use it to show my items on models. I think it’s fantastic & saves me so much time. It’s great for tracking trends - and suggesting new items/slogans/designs.

If a customer has an issue (usually with international deliveries!) I screenshot their message & ask ai to respond. Because I’ve used it so long it typically replies how I would - same tone, etc - but it takes the emotion out, which is really helpful & stops situations escalating.

Overall I find it really helpful, it allows me to spend more time on my orders, rather than pissing about with admin - which I hate 😩 I don’t use it much for personal stuff - though recently my DH had an issue at work & I used it for that. Like I said, it’s great for taking the emotion out of situations which is great for a hot-head like me!

You do need to remind it to be brutally honest sometimes or it’ll cheerlead you on when you’re clearly being stupid - but other than that I’m a big fan (and a subscriber!)

You can't manage to write a basic email or message anymore? Oh dear. No, no, most of us can still write emails and messages. I use it about 2-3 times a week to find a citation, or a recipe perhaps.

And you Pay?! For AI?!

Good lord, no.

I don't think for a lot of people it's just that you get crap out of it. For some people it's far more fundamental than that.

AI is environmentally horrible. It's taken over so many parts of the internet so as to render them unusable. It's removing the need for people to think for themselves, to research or critically evaluate things. It's already costing people their jobs and their livihoods and there is nothing the ordinary person on the street can do to stop that.

We're basically sleepwalking into an AI wilderness because a few big companies stand to make a mint out of it.

Don't get me wrong, I know that there are benefits from AI too, but we need to (or rather needed to, because it's probably too late now) pause and think about what kind of world we want, and legislate to control AI accordingly.

Right. I can and do use it well when I want to. I occasionally use it to edit one of my own images for a site I am on, find a citation or check a recipe - it's actually not difficult (at all) to use ai and learn how to use it well.

And there's nothing remotely defensive in pointing out that if you're relying on ai to the exent the OP is you are damaging your ability to think for yourself.

Give it a test.
Put your user name in and a forum you regularly contribute to...Mumsnet will do.

I did this (not using this forum) and some of the info it gave about me was completely false.

The info it takes in is not always the same as the info it gives out.