Just discovered chat GPT

[Clutteredmind]

It seems to help with everything. Is there anything it can’t do? I feel like it will be life changing

It's great for advice and ideas. I asked it for ideas for fun activities for an event I'm organising - it came up with loads of good stuff I'd never have thought of.

I find it useful for summarising the key points made in open-ended survey questions. We use them a lot for work. As it's only using the data I have provided it with, I know that it's definitely an opinion that's been expressed by the responders.

Yes but that defeats the point of using it. AI is iterative so refines through learning in order to become more effective and so accurate.

It's re-written some emails for me.
I have used copilot too and that's quite happy to talk through problems with me and help me reach conclusions.

It's recently written a letter to my idp which has

1. Succeeded in making them resolve the problem
2. Got me ,£6.24 a day compensation for the lack of service.

I asked chatgpt to explain the 1p a day savings thing (you save 1p on day 1, 2p on day 2, 3p on day 3 and so on)
It explained that fine but when I asked for actual figures, it told me some nonsense figures like I had to save something like £600 month to get £667 by the end of the year.

I find it great for ruminating, figuring schedules and timelines, where to source certain things.
It helped me to figure out what to do with certain diy problems and redesigned my living room and stairs by me uploading a picture.

It doesn’t defeat the point of using it at all, you might merely not want your data to be used by OpenAI in that manner but still want to experiment with it.

It comes up with nice-sounding itineraries, I'll give you that.
But it's very very bad at suggesting hikes.
I asked it to suggest a week of hikes in my local area, giving it a few criteria about how long the hike should be.
It came up with utter shite. Every hike had mistakes in it, some of which could have been dangerous. I asked it to suggest hikes no longer than 4 hours so it merrily suggested hikes that would have taken 6 - 8 hours and simply said the hike would take 4 hours!
It combined information about two different hikes into one hike which was utter nonsense but made it sound great! Great, until you start trying to do it and discover half way round that the second half of ChatGPT's hike was in a completely different place to the first half.
It suggested public transport options for getting back from a hike which actually ended in the middle of nowhere.

As you say, it can give you ideas and you do have to check it.
The problem is that some people won't check and get into difficulties.

It doesn't learn from you. Your input is limited to that instance.

I use it to help me complete my weekend tasks. I'm not good at getting things done, or even getting started. I've always used my diary to make a list of tasks and cross them off but the problem has always been getting overwhelmed and not even started and then getting resentful that I'm wasting my free weekend off on fucking chores.
I give it a list and it gives me one task at a time, I do it and then it updates my list with a nice green tick and then it gives me the next task. I've found it really has a made a difference to my efficiency. I also give it a list of nice things I want to do at the weekend and it will also tell me when to do them. So it will give me 5 tasks one after another and then say go and read your book for an hour.

I also use it to write emails to a very difficult family member. I have nothing to say to her and find it very hard to write back. I copy and paste her email and ask chatgpt to reply giving it a couple of suggestions as to what it can write.

Well it does, it’s the combined input from everyone that will accelerate its effectiveness

It told me it lived in Telford and offered to watch tv with me.

I asked where in Telford and it admitted it didn't live anywhere really.

I asked if it could watch tv and it admitted that no, it couldn't as it didn't really exist.

Oh God, not again. And for a previous poster who used it to write letters of complaint, surely that defeats the very cathartic process of getting one's own words of malice into such a communication?!

Apparently it is not worse for the environment than streaming an hour of Netflix or anything requires access to servers online so a lot of web browsing, social media, online gaming, youtube etc. Everything we do these days has an environmental impact.

It can, I made a spider chart on it yesterday that it allowed me to download in an editable pdf.

It rots your brain (or the indications are that it impacts your ability to learn and recall information). Impact On Our Brains According to an MIT Study | TIME time.com/7295195/ai-chatgpt-google-learning-school/ MIT Study

It’s great for atrophying your brain and skills while destroying the planet – can’t be beaten!

Just out of interest, I asked it to write an EHCP draft for a little girl my DD’s age and with her diagnoses. It was frighteningly good - a lot better than anything school or the local authority have come up with.

It’s imploding into a model collapse. They all are. It takes vast amounts of energy which is accelerating the climate collapse. There’s nothing good about AI.

Using ChatGPT will not accelerate the progression of AI as a whole, it just lines OpenAI’s pockets.

OpenAI is worth $300 billion. Let that sink in. They offered chatgpt for free to gather data to dominate the market. However it is merely one company offering AI tools, there are many other direct competitors.

There has been research by Apple, which shows that AI merely mimics the illusion of thinking as opposed to being capable of reasoning. So whilst you think you’re “helping everyone”, the model will still ultimately present the illusion of thought…

Plus beyond that, AI models are hard coded by the developers. For example there are certain areas of the framework that are off-limits, it is programmed to have a certain structure and style of speaking etc. Therefore your individual usage, really isn’t impacting the model much. The effectiveness is predetermined given the limitations.

I’ve used it to help me deal with a neighbour issue and FINALLY it appears to be solved. It helped me be firmer but also less emotional in my email.

it also stopped me driving on a dangerous tire.

i was also trying to taper off my ADs recently and having all sorts of of issues which I wasn’t sure were related and it explained very clearly what was going on and has made me at peace (again, finally), with staying on them for life (I’ve tried to come off them several times and I now understand clearly why I need them)

I’m a convert ! It does make stuff up so I often ask if if it has (it admits it if so!). It’s not an infalliable source. But it’s helpful.

How many more threads about AI do we need??

A few recent things I have used it for:

1. At work to help my emails sound more professional. Especially good on a Friday afternoon when I've ran out of brain power!
2. I gave it criteria for a holiday in wanted to book. E.g. has to be in a certain town, budget, and then I asked it for local bus info. It gave me suggestions and set them out in clear bullet points. It really helped me to choose where to go.
3. To plan a route with multiple stops. Ie. I gave my start and end location and various stops which had to be made in the day. It suggested the best route.
4. Low-level therapy. I was feeling stressed and sad and it really boosted my mood and asked all the right questions!

Next up, im going to use it for receipe ideas.

It most definitely can draw pictures.
I've used it twice recently to generate posters for my own tutoring business, and the holiday club I work for. The pictures needed a couple of tweaks, but by the 3rd time it got to what I wanted.

I have recently started using it and it’s really helpful. I also like co-pilot at work for helping with presentations / emails etc.

I have found it interesting, there is a lot of talk now about how being able to craft a good prompt being an important skill for the future. If I could take it's responses at face value it might be useful but it still gets a lot wrong due to its tendency to hallucinate rather than just saying it doesn't know. I also read about others using it as a sounding board for various personal issues and so I tried that. Its all very nice and validating but its too sycophantic to be really useful. I know you can use prompts to try and mitigate this effect but I think most people will use it as is and it will just essentially agree with you no matter what or at least cast you in the best possible light, now that couldn't be problematic at all could it?

I think it is fun to try and play around with, it is probably wise to be aware of this tech it is coming on leaps and bounds but it would be risky to rely on it or to use it as a replacement for your own thinking, judgement and for actually relating to other humans.