How are people using AI to help with everyday life?

[Question12458]

How is your household utilising AI to your benefit? Looking for some inspo

I’ve been using ChatGPT (paid) for 2 years, this week I spoke to somebody who’s got a 2nd brain. (Collects all data from multiple sources) and is totally customised to them. Totally opened my eyes to what we can achieve

I take it none of you drive cars, go on aeroplanes, have a mobile phone or DC either?

Great - and by inputting your unique personal thoughts into the LLM you've made them available for anyone who asks the LLM for answers next time!

Because all technology is equal, right?

"I take it none of you use sarin gas, pervert glasses or assault rifles, either?

That depends on what you are using and what settings you have chosen. Copilot is governed by the Microsoft data protection policy and cannot use your input for machine learning.

https://learn.microsoft.com/en-us/microsoft-365/copilot/microsoft-365-copilot-privacy

Why are you so emotional about the uses other people find for AI?

It’s absolutely standard across many industries, not specific to my employer. You’re right about the job cutting, of course.

Quote " I think a lot of people dunt know how to use it and are scared of it."

I don't think it's a matter of not knowing how to use it, I think going by your list above, that people don't need to use it.

Genuine question. And also, the same question to so many here. Do you know how to use the normal internet ? I am sure you do, because you are here.

Take a UK trip for example. Say you fancied a weekend in Whitby. Would you not just search for Whitby on the web ? Maybe use Tripadvisor etc ? Maybe look at a paper map to find the best route ?

I just don't see anything on your list that I would need to use AI for. Indeed, I would not want to use AI for some of those things because AI is not 100% accurate.

Yes, I have tried many different AI models that are we are encouraged to use at work.

The most impressive thing they do is transcribe meetings and calls with... 95%? accuracy.Their summaries are OK but they will sometimes hallucinate The moment they try and conclude anything about what's going on, they take the most route one, generic interpretation, although if you prompt them with alternative ideas, they are happy to pretend they had also thought of that, and that it is the best choice.

They don't count or understand time very well, or make maps.

Image and moving image generation is usually good, as long as you start with quality original inputs . (i.e. not those terrible AI-slop charity event posters you see everywhere.)

They are really quick to churn out variants of a thing - remixing things you've done before. So you can get them to sing bad power ballards or make boring podcasts or create generic lesson plans about the Vikings.

But actually getting them to create or critique something new and you get something that looks OK at first glance but either falls apart under scrutiny or is the most basic generic take pumped up with fancy language.

I use it to help with treatment options at the doctors having been fobbed off for years. I put all my symptoms and variables in and it gives me options to discuss with my doctor that I wouldn’t have known were available.

This is one of the reasons people think AI companies are overvalued. People don’t need to use LLMs for 95% of tasks- they could as easily use SLMs more cheaply and with less waste of resources.

But I think your distinction between using AI and using the internet misses something. When you do a google search, Google uses AI to interpret your question, rank the results etc. I’m not talking here about the annoying AI answer at the top- basic search results are provided to you with use of AI. Everyone who uses the internet uses AI whether they know it or not.

I think so many of these use cases must be for really thick or incurious people.

Honestly, who needs an AI to suggest what to eat from the food in their fridge or what to wear from the clothes in their wardrobe?

If I want a weekend break within 3 hours drive, I look at a map. It's really easy.

And if I asked an AI, and it suggested five places, I'd want to look at a map to check where they were. And I wouldn't want to read the dead bullet-pointed prose of an LLM telling me about them, especially because, as you say, they make stuff up.

I am emotional about other people's uses of AI when I see other people still believing in the integrity of the big company's data protection policies after the scandals and breaches that have occured over decades.

(And I believe the protections you cite are only for Microsoft enterprise clients, anyway )

Hmmm, I must say my experience has been very different! So much so that I secretly wonder if I got routed to some poor human specialist who is shackled there being fed gruel while s/he trains the LLM with personalized input to my queries 🤭😁 If so, they certainly type fast!

We have autotranscription on for work calls. It would be really useful if it worked. Unfortunately it doesn't really work well at all. It struggles to parse 'non-standard' accents (because voice recognition software is largely trained on RP or Standard American) and will sometimes get important details wrong as a result. If you are working in an area where you are dealing with people with distinctive strong regional accents (including use of slang) or who might not have English as a first language it's basically useless.

I am Scottish. I defeat all voice recognition :-)

That's exactly what people say about their horoscopes. We all know that experiment, where people reported remarkably accurate specific insight about their personalities supplied by an astrologer. And actually, everyone got the same generic horoscope but read into it their own circumstances.

It's just a large language model. It has no insight into your creativity. It is just regurgitating generic feedback aligned to your subject matter or style.

Honestly, I think generative AI is for dimwits. Don't be a dimwit.

I mostly use it for recipe advice - eg I have a chicken breast, pepper and tin of lentils, what can i make tonight?

I don't think I use it for anything that I couldn't Google, use books, or think for myself. It's just a lot faster. And it tells me what I want to hear.

Hmm, you do know I was joking? don't you?

I'm not a dimwit, I've worked successfully as a professional editor for 30-odd years, and find AI an extremely useful tool when refining my own writing. I could share the useful critiques I've had, but I'm not going to compromise my privacy just to convince you to stop insulting me 🤷🏻‍♀️

I asked a family history question, and recoiled at the unverified slop it threw out. I find it terrifying that so many people will take information without triangulation then corrupt ancestry/ other sites.

I'm a software engineer, so AI is just an inevitable part of my job and life. My husband however was a pretty senior head of dept at an advertising agency and decided a few months ago that he is tired of shitty management, long hours and reduced staff with higher workloads so he took 3 months of stress leave to figure things out. He's always been into finance and investing and that's where his passion has been for a while so he used those 3 months to work on that. With the aide of AI, he has built his own firm of autonomous agents who give him insights, documents, research analysis and help him use a logical approach to investing vs an emotional one. AI is incredible at crunching data and finding patterns, so analysing financial documents is significantly easier to do with it. His salary before was quite high so his taxes were also idiotic and he has made more with this than he was when working, month on month with some consistency, while paying significantly lower taxes (always a win). He's now decided to take voluntary redundancy with a full year payout and plans to make investing, with the help of AI, his full time job. I've never seen him so happy.

Or eat meat. People love complaining about the water consumption of data centres yet animal agriculture consumes 200 times more freshwater than all data centres and AI operations combined.

People just like complaining about the things they cannot change while doing fuck all about the things they actually can.

Yes it is a LLM. What it is remarkably good at is analysing large quantities of data to find patterns and to identify information that’s of use to your specific question.

In this sense it’s only as good as the prompts put in. It’s also necessary to have someone who understands the topic to sense check the information coming out.

But it is incredibly good at certain things and has already significantly changed some aspects of my job. That’s the problem to be honest. If it wasn’t actually that much use, employers would not be embracing it.

I was so unimpressed with early versions that I stopped paying attention.

It leapt to my attention a few months back, when I had a problem installing Ubuntu Linux onto a new Acer laptop with the Windows operative system preinstalled. A series of infuriating problems had me stymied until AI explained that Windows was "protecting" the drive to prevent me from installing Unbuntu on a dual boot system, and Acer had decided that my attempts to change the BIOS settings were malicious so it removed my admin privileges and made it impossible to create an admin account. It took 7 conversations, where AI fed me 6 different solutions that didn't work before it mentioned that the only way to regain admin privileges on this make of ACER laptop was via a secret series of keystrokes. Finally I was able to access the bios, switch off the disk "protection" and get the job done. I'd expected it to take a couple of hours, and instead I wasted at least 18 hours on it, 8 hours of trying all my own tricks, then another 10 hours spread over several days following instructions from Google AI, Brave AI and Copilot before one of them revealed the secret keystrokes that ACER like to keep to themselves.

After reading this thread I asked Google AI about filing a tax return for the year 2025 - 2026 and it told me that I had until midnight tonight to set up a digital account with the tax office so that I could have until 31st January next year to file a digital tax return, or else I'd need to ensure that all the paper files were sent to HMRC to arrive before the end of October.

So I just spent 4 hours trying to create a digital account. I had one that I set up in 2004 after moving back to the UK, but it was with an email address and phone number that I can no longer access. I'm impressed that the UK manages to make their tax return process even more baffling than the Swedes, who I thought were the acknowledged world experts in tax mystification.

Now I'm wondering if AI will be able to talk me through the process of filing a digital tax return that takes into account the tax already paid at source on my Swedish old age pension, so I can work out if I am liable to pay tax for that year or not. If I am it'll only be a very small sum, so I'm not too keen on forking out to get help from a tax accountant.

But my head hurts from the ever increasing levels of bafflement it was exposed to trying to work out what the strange words in the questions actually mean. Even clicking on the explanations failed to explain them properly. I wonder if I should download the paper forms just to get an idea of what they ask before committing to answering yes or no to a load of confusing gobbledygook that might end with me being charged with filing a fraudulent tax return if I guess wrong.

Again, I am impressed, I had no idea the UK was able to make their system even more confusing than the one the Swedes are tormented by.

I use duck duck go as my browser and search engine and was able to turn off AI in their settings. I was able to turn it off on my mail app too.

I use AI quite a bit. I use alexa to add stuff to the shopping list as I am using the last of it, announces dinner time, set timers and alarms. Also most of our lighting, all lamps are on Alexa and a few set to come on at sunset so if we are out the dog is not sat in the dark and house looks active. I use chatgpt to help organise days out or plan trips and help me find the cheapest rates etc. Also asked for help choosing an outfit for a party next week. It searched multiple website, some I had never thought to look on and links to all and ways to accessories each outfit too. Meal plans and writting me a shopping list to go woth it. Also used it to find all local 6thform open evenings near me to save me looking on each website. Help with revision schedules for my daughters upcoming GCSEs and links to free helpful resources. Also use copilot at work. Mostly if I need to send something more formal to just check over the coherence and sometimes to change my tone if the email sounds too aggressive. Also to pop reminders into my calendar of colleagues I need to speak with and checking each of their diaries to find times that wont conflict. I find it all extremely helpful.

A lot of the time, a "the kind of things people might say next in this kind of situation" machine is good enough.

If it asks "why" type questions and plays back what you say with a different emphasis, that's going to trigger your own thoughts and refinements. Throw in a bit of pattern matching to bring in possible connections from the outside world you hadn't thought of and you have a pretty good sounding board.

Half the challenge of creativity is to think outside your own box - plenty of artists have used completely random techniques to jump their own creativity onto new rails. The creativity doesn't need to come from the AI, the AI just needs to be a trigger for a thought you would not otherwise have had.

(80% plus of what humans output is standard follow-the-crowd stuff anyway. Most jobs don't want creativity, they want someone who understands the problem well enough to pick an appropriate response/action from a well understood and resassuring toolbox. This is why junior consultants are being replaced by AI!)