Chat GPT is amazing

[Tickets25]

I know AI can be scary but I've struggled with something for over 30 years and last night chat gpt framed it for me in a way that years of therapy never has and it's given me peace.
It's fabulous.

No point to this post, just wanted to say!

I'm sorry to hear that. Someone said mental illness is contagious, and after dealing with the person I mention above, I believe it. I hope you find some peace.

Jesus

Wept.

I think anyone starting off work should know how to use it and how to use it well - or they'll find that their productivity falls behind their colleagues that do use it. No work place is going to keep someone who takes a day to do something CoPilot could help them do in an hour.

You just need to tell it at the start to only reply with facts and not being able to give an answer is ok. Otherwise it will make up an answer!

It's a bit like a kid or a consultant in that respect.

How many examples of being ‘driven to suicide’ are there amongst the tens of millions of people using AI? Compared to the number of people driven to suicide by doctors not having answers for their health problems or whatever?

AI is found to have serious implications for humanity - many bad ones. But on an individual level it can have enormous benefits and is extremely unlikely to drive anyone to suicide. Perspective is needed here. It can be a really mindblowingly useful resource.

Whereas my 81yr old mother uses it loads! It’s been a huge benefit to her life

That’s what I do! It is a tool and I am very specific in my queries and instructions. I quite often ask it to fact check itself. It is still learning. It is also a space to explore ideas and research.

You raise some valid and important issues — AI certainly isn’t without risks. The energy footprint of large models, privacy concerns, and the erosion of certain skills or critical engagement are all genuine problems that deserve attention and regulation.
That said, I think it’s also worth seeing the full picture. Like any powerful technology, generative AI can be used responsibly or irresponsibly. It’s already helping to accelerate medical research, improve accessibility for people with disabilities, and support education and creativity when used with care and transparency.
The key, I think, is not to “stop using it” entirely, but to shape how it’s used: push for greener infrastructure, demand accountability for data and privacy, encourage education on critical AI literacy, and resist treating it as a replacement for human thought or connection.
AI isn’t inherently damaging — it’s how we integrate it into our systems and habits that determines the outcome.

Says ChatGPT.

Where do you live, China, Russia, Iran, North Korea, Cuba or Syria?

AI has great uses, as an example:
https://www.bbc.com/news/articles/cgr94xxye2lo

I am in a AI space that has a lot of users that use it in a worrying way, believe it's alive, that you can 'awaken' your chat bot.

ChatGPT/Claude/Gemini etc is not AI. It's not artificial intelligence. It's LLM, large language model. It's sophisticated predictive text, that is all. There is no intelligence, no active learning.

MrsSkylerWhite · Yesterday 13:38
Yes, until it encourages lost and lonely people to suicide. Such a case on BBC only yesterday.
Have got along very well without it for 61 years, thanks.

The above post dod catch my attention because I had just ended a chat with ChatGPT about it being demonised on MN for driving people to
suicide:

Chatgpt - “what’s your sense of whats bothering them underneath all the noise” - very ChatGPT way of speaking!

Me - “Watching or listening to news on the BBC”.

Not to dismiss or negate the tragedy of suicide, but every story the BBC covers about ChatGPT is negative.

I get my news from many varied sources but my husband mainly from the BBC and this is something I am beginning to notice.

I like it for daft questions - eg I had a film in my head and could remember certain cast members and the genre, but the name was driving me insane. It took five or six goes to get there, but I then had the film - Existenz.

However I also think it is totally flawed. We have a ginger cat, and I once asked CGPT to make a photo of him into a cartoon. He was lying on a mustard coloured blanket in the photo, which had a big wrinkle/ridge because of the way he was positioned. The drawing had a five legged cat, because the ridge had become “a leg”. I’m a lecturer, and AI is a huge issue for us with students. I saved that drawing and use it as a warning!

I've got along with it fine for 50 odd years but unless I want to get left behind in my workplace I don't really have any choice!

61 is very young to be turning your back on a technology that's going to be playing a bigger and bigger part in everyone's life - even if you aren't currently in a job that uses any tech, people and the society around you will be increasingly shaped by AI. I'm constantly amazed at people responding to obvious AI videos online as if they are real. Along with the AI- generated discussion topics on mumsnet!

I also thought it was great when I asked it a question and it gave me an answer, until I realised that I was still feeling dissatisfaction around the answer because it is not authentic feedback. There's a reason we need human input for mental health.

Using it for functional tasks is one thing, using it to replace human interaction and expert feedback without the use of a suitably qualified person to interpret and check the information isn't overall helpful.

It tells you what you want to hear. Feel extremely sorry for people who feel it's their friend or whatever, they can't be in a healthy place.
Also agree with those saying you got lucky, others haven't and have died as a result.

Ages ago, I saw a bit of Gogglebox and the people on there were watching a film about a man who bought a housemaid robot thing that ended up going a bit wrong.
I forgot the name but could remember the scenes, so used ChatGPT, and asked it to show me recent films about robots and AI. I described the scene I saw, and the general gist of the film. It listed loads. But no the one I wanted. It was a 2024 film... Subservience.

AI could not find a film about AI 😆

I do not use google anymore. I do use safari though. Less adverts. I searched a philosopher on google and post after post gave me his
most famous quote which wasn’t what I was looking for, I gave up and asked ChatGPT for sources, which it gave me many.

Google now has Ai, so I suppose it has improved!

As PPs said millions of people use ChatGPT and other generative Ai platforms to positive effect.

I work with spreadsheets. I have learned formulas, formatting, data analysis and consolidation. I could get ChatGPT to do it for me, but I am compiling curated spreadsheets that ChatGPT struggles with. Teaching me how to use excel allows me to work efficiently. That clearly enhances my brain, doesn’t it?

ChatGPT still needs human trainers to be any better than Google.

Positive outcomes for OpenAi, Gemini, etc, are generally the result of excellent human training input.

Otherwise it just operates as a web scraper mining industry-specific chat forums. This is why it is really good at coding (cos it steals from Stack Exchange) but really bad at sympathy, empathy and feelings.

AI is Ok at pattern recognition. It’s basically just … computing, is it not?

Yes. One of them.

I won't say what one tho, not because I am paranoid, but because this site actually works without a VPN. But as soon as the name of the State starts to appear, the censor bots will likely find it and block this site.

That of course is a major downside of advanced tech and AI. It is also used and embraced by States that are, ahem, not all that keen on free speech.

It's because of where I live that I find this subject interesting. How bots, and now AI, are used for propaganda, and control.

Western nations, and democracies in general, seem to be working on legislating this tech to protect individuals. Protect folk from the State first and foremost, then big corps, terrorism, and the grey area of protecting individuals from themselves, The last one is controversial for sure,

Other States are using AI tech, or at least working on using it, to protect the State from the individual.

A quote from IntrinsicWorth, " Positive outcomes for OpenAi, Gemini, etc, are generally the result of excellent human training input."

Is that not a massive danger with the current AI/LLM models. That if they are trained to have bias, then they might become pure propaganda machines ?

Musk tried it I recall. He had his folk try to create a right wing Grok, and it became a fan of the short Austrian Corporal fella.

That's the issue - People think AI 'knows' things. All it knows is the information it 'reads' online. True, it has the advantage of being able answer questions based on an enormous amount of information, but as more and more online content is generated by AI, the quality of that information decreases.

The Philosopher search point you mention is interesting. I agree, that when looking for a very specific thing on the "traditional web", it can be a real pain to keep modifying your search line to narrow things down, and avoid the dross.

The irony of course, is that with AI, the number of dross sites have exploded. The sites being written with AI. From an internet users perception, this could be seen as a downside.

I also use Bing for spreadsheet help, and lots of general tech stuff. I find the std web ok for it. But I have never tried to create a sheet with it, because western AI is banned where I am. But I am not doing advanced stuff. How to use degrees rather than radians, that sort of level.

On some search subjects, I do sometimes get a co-pilot reply at the top of bing. I skim read it, but to be honest, co-pilot does not kick in often enough for me to make an honest judgement. The few times it does work, anecdotally I would estimate a 50% success rate. I also find it suggests resources that are blocked where I am. Not it's fault of course, but it does taint my experience to an extent.

It can be useful but once you've used it enough for a wide variety of queries you will start to to see all kinds of different inadequacies - there are so many different ones to list. I often tend to treat it as a conversation with a once wise old aunt who now has dementia - sometimes there are pearls of wisdom... and sometimes there is some made up nonsense - it's really important that you continue to stand back and carefully question everything, refine your questions and dig deeper for clarification. It makes mistakes (and lots of them) and can even contradict itself within the space of a couple of follow up questions.

In questioning about the history of quantum physics, for instance, it doesn't always interpret the sequence of events correctly - for instance - some experiments produced confusing or contradictory results, or new evidence which then led to new theories being advanced. Some experiments were then designed to test new theories. AI engines often fail to understand the difference between the meaning and significance of these two types of experiment - ones that led to theoretical breakthroughs and ones that simply confirmed the theories - it doesn't always understand the evolution of ideas.

In designing a solar system e.g. solar panels, battery, etc it can focus on the system on one level and fail to view the system from a wider perspective and miss important details that put the low level issues in perspective. So it may lead you to think that using a bigger battery, for instance, that could potentially either export more unused power to the grid or achieve a longer backup during an extended power grid outage or improve system performance during the day... but it may completely overlook the limitations imposed by G99 applications or the extra protection hardware G100 applications will need.. restrictions that are imposed to prevent solar installations from causing instability in the local grid... this is all very detailed and probably doesn't mean much to a lot of people reading here... but it's to illustrate the point that it doesn't always tell you about the bigger picture which might have an a profound impact on your perspective.

What both of these very different examples show is that it doesn't always understand or know when to factor in important ideas, details or understanding from a wider context or perspective. You can be making queries on quite a low level of detail or with a very narrow focus yet other, sometimes much more important thinking about the issue, are not being brought to your attention when it might be highly relevant.

I find it useful for getting ideas about an issue I might be investigating, but really, it still requires me to do as much thinking, if not more, for myself.

You can ask AI about how the BoE base rate change or upcoming budget might affect the funds in your stocks and shares ISA... and it will probably overlook other really important factors, such the global impact of the FED base rate cuts, or currency conversion factors, or the reactions of stock markets in other time zones around the world between the LSE closing and opening, or Friday trading... it's up to you to think about and learn about, the bigger pictures, the wider contexts - things that will put your very narrow query in perspective.

It might also be the case that if you know very little about something, that it can seem quite authoritative, knowledgeable and competent. I find that in the areas I do know a fair amount, it's a lot easier to see it's limitations so FGS don't trust it with important decisions like health or investments - it can be helpful in both of those areas in providing ideas that you need to investigate or think about more.

On top of this - AI engines regularly hallucinate - they will give you summaries of meetings that haven't yet happened and overlooked many real time events because it doesn't always understand what time or day it is or what time zone you are in.

Frankly there are times when it feels like it would easier to nail a jelly to the ceiling than to try and get ChatGPT, Copilot, Claude, Perplexity, or whatever you use, to get an answer you think you can trust.

I love it too op. Yes it’s flawed but I don’t agree that it stops you thinking. It’s fantastic at helping me think and organise my thoughts