What do you use AI for?

[thehousewiththesagegreensofa]

I feel as though I need to take a leap into the unknown and embrace AI. But I'm not entirely sure how. My mind was blown the other day when I read on a thread that someone was using ChatGPT to generate recipes for her and all she has to do was provide the suggested ingredients. That just hadn't occurred to me as something you could use it for. So, how do you use it?

I use Microsoft Copilot.

Today, Copilot and I drafted all of my 2025 plans in under four hours (I am a CEO of a Tech Company).

Business plan
Annual report
Quarterly Board Report
Impact plan
Marketing plan
Sales plan
Finance plan
Operations plans x 3
Customer excellence plan
HR plan

I use it to help me draft emails.
I use it to take meeting minutes and record actions.
I use it to research trends
I use it to support my teams performance and learning needs

It is the best thing ever.

I have created Pixar style film posters of my dog and I have sworn at it to see how it would react (can’t remember how it reacted though, so not really worth it tbh). The Pixar style film posters were quite cute though (the writing on them is shocking but the images are fun) I used the bing one for that.

Having just gone through a recruitment process I’d advise against using it for cv writing. It removes any personality and meant a large number all read the same. I had 49 applications to read through and more than half I’d guess were ChatGPT.

Try to avoid using AI for anything that you're not good at yourself, unless you really don't need that skillset in life.

Eg if you're a poor writer, and you now get it to do all your drafting, your writing muscle will only deteriorate and before you know it, you'll be having to use ChatGPT to write your birthday card messages.

Your brain is a muscle as much as your body. If you delegate intellectual processes to the machines, eventually your brain will become flabby and less high performing. Use AI as an enhancer, not a replacement, for your thinking.

Personality, creativity, critical thinking etc are going to become more and more important attributes for humans to develop. Not just for jobs but for maintaining our humanity.

Interesting thread. I am wondering in 1--15 years there will be less employment for office based (or home based) 'white collar' type of jobs and manual labour will be higher paid. Difficult to advise dc in terms of career path, uni etc.

I saw this ad recently. I think building skills will definitely be in demand (or at least builders working for their AI overlords who ensure they are efficient!)

Gen-AI, is making human personalities more homogeneous.

Eg AI-powered social media algorithms which seem to shrink diversity of thought and expression, as well as confidence to express individuality.

Helping your children keep their individuality (in thought, expression, actions) is going to be really important as ChatGPT etc influences humans to flatline, question less, accept bland answers, and become ever more alike. You already see it happening in politics, where personality trumps skills and experience. I think we'll increasingly see it at the social and community levels too.

I'm interested in the way people are talking about using it to help explain maths. DS is terrible at maths. I am terrible at attempting to explain it. Some of the videos he watches do not help at all. But perhaps we can experiement with HIM asking chatgbt the question in his own way, to see if it then generates an answer that works for him.... This is very interesting. I'm going to suggest it to him.

I work in communications. ChatGBT is extraordinarily helpful BUT like anything, it depends on the quality of what you input. So, if I ask it for a 750 word blog on subject x, what I get back is frankly completely pointless. If I ask it for a 750 word blog on subject x, covering these 3 key themes and highlighting the importance of x,y,z and to do so in a tone that is business informal as written by the CTO of a fintech organisation.... I get something much much better that then I can edit.

Having said that, I prefer to write a rough first draft myself as that way I have all the concepts. Then I put it through ChatGBT. Then I go through again and mostly remove the ChatGBT changes but incorporate a few, with edits. often this is particularly helpful to more quickly create something that flows better.

I often use it for technical concepts I need to better understand. I can google, and do, but I get a more specific answer with ChatGbT.

And most importantly, I use it for HEADINGS. I am terrible at headings. ChatGBT does have an annoying style of heading, but you can ask it to edit it and then usually between the three options it generates, I can create a heading myself. This has been life changing for me.

Ha!

All this reminds me of the debates about using social media in workplaces and professions.

It was either going to be the magical answer to everything or bring about the end of life as we know it.

When in fact it was neither. It's. Just a comms tool.

That's all AI is. A tech tool. It will save some time, resource and build capacity.

Lots of people will bury their heads in the sand, sneer, put things in the bin

Others will embrace it and leave them behind.

I don't use it at all.

What's its carbon footprint for all these little tasks people are using it for?!

Climate change is threatening us with extinction and huge server farms are gobbling electricity running these AIs... just so people can ask for recipes and word emails. I feel like that's the moral equivalent of driving an SUV round the city.

I so agree. I am old enough to remember people being horrified at the use of computers for word processing and email and complaining that we would become careless and stop paying attention and would forget how to spell if we used spell check etc.

Instead, communication is improved and people can focus their time on actual value add tasks. Plus, the thing everyone was concerned about is NOT the problem that actually came up - which, in my opinion, is the shift to being "always on" and the expectation of artificailly high levels of productivity.

From what I’ve read each query is about the same carbon usage as posting a tweet (or a post on a forum…)

Can someone explain in layman terms how this actually works? Who's inputting all this knowledge into the data and thinking of every scenario that can be thought up?

I put some random waffle about how I'm emotionally drained with disability services, and it actually gave me something I hadn't thought of!

It is freaky

Can someone explain in layman terms how this actually works? Who's inputting all this knowledge into the data and thinking of every scenario that can be thought up?

Depends on which particular AI ... some are being trained on existing internet sources so I'd guess quite a bit is regurgitated Wikipedia. The input may not be reliable and won't necessarily be up to date.

I don't believe people are 'thinking of every scenario that can be thought up'

I don't use it. It isn't something I need or particularly want in my life.

I accept some people can use it 'for good' and that's great. It's not really for me though.

I feel like humans are disconnected from one another and reality enough as it is. I'm not sure we need this to help us in that endeavour.

I use it for writing code I haven't got time to write or if I'm struggling to get something to work. It always needs tidying up and a bit of rework so you still need to know what you're doing, but it's a great time saver.

but you then have to look through various articles which may or may not be accurate. AI is not 100% trustworthy but instead of giving you a list of links it will in a few seconds write you a concise explanation. Meaning you don't have to look at a range of different other links to come to the same conclusion.

See I don’t think that communication has necessarily improved. Yes we can email, but all the pointless time spent emailing, asking things instead of working them out, sending pointless tweets. If anything, all this communication batting backwards and forwards prevents focus on specific tasks. Social media is basically devoid of any true value - it’s just noise.

Plus I do think loads of people are terrible at spelling and grammar because of spellcheck!

AI is going to automate almost all professional jobs. It's only in its infancy now but soon it will compete with doctors (not surgeons/dentists), architects, researchers, designers, writers, lecturers, administrators, PAs, counsellors, market research analysts, financial analysts, personal financial advisors.

Law is relatively safe as is structural engineering, politics, physiotherapists, gardening, musicians, construction, anything that requires human physical skill to do the job, as robotics are nowhere near where AI is. Carers, primary school teachers, nursery nurses and actual nurses should also be safe .There will be new jobs related to AI management.

Only time i use is to get excel formulas or to do the grunt work of turning a long list into something separated by or so i can use it in a search at work...

Even reading the thread, half of it feels like it's in a foreign language!🤣

AI learns on the data that is inputted. Much of this may well be data already on the internet, but it will also be receiving specific data sets that may or not be available to the rest of us. For example, I would expect that chatgbt has, over time, had lots of medical and financial data inputted.

In terms of answering questions you haven't thought of yet - that's where AI is so smart. Alongside learning all this knowledge, it also learns language so it can understand the question being asked, and then based on that, figure out the best way to answer the question. This is what makes it different to say google which is looking for search terms only - increasingly, google and similar are also using some AI to improve search functionality, but true AI goes that much further.

This is why the data that is used to teach AI is so important. For example, some early examples of problems come when the data is, itself, flawed. For example, there's a lot of implicit bias in data sets around race, gender and sexuality and it is taking significant effort to stop this from impacting AI. Not always effectively.

To add to the above, large language models (which is what ChatGPT etc are) will also train on their own data. This is why you need to ensure there’s human oversight because if an error or bias is present, it will only get further reinforced.

Thats what the poster above is saying anyway, I’m not contradicting but just to make it really clear that the model is continually training and refining (although those refinements won’t be available to the public in real time, I think the free tier of ChatGPT dates back to 2022) on ever greater amounts of data. And there’s ethical concerns there as well with how that data is gathered, stored, processed etc. AI ethics is a very interesting area.

They are large language models. The reason they’re called this is because they’re trained on massive amounts of content scraped from the web. Basically, it has been fed enough training data to recognise human text input and predict the correct answer for your query, based on prior learnings. They use deep learning, which is an analysis of how words/phrases etc fit together. They are also prompt tuned (fine tuning) to particular tasks, like writing code or recipes.

im leaving a lot of information out there but I hope that helps! LLMs are a real rabbit hole topic.