Artificial Intelligence

[Leon_C]

I find this board to be a reasonable cross-section of the general public - so, on a good day, worthy of a quick bit of off-topic research.

So - I put it to you good citizens of UTC...

AI is a big area of development in my business (we’re a digital marketing & web agency getting increasingly involved in AI workflows). Because of this, I tend to circulate among people who already know a bit—which can create a bit of a bubble.

So I’m really curious: what does AI uptake look like outside of that bubble?

What’s your general awareness and usage of AI at the moment?

Have you tried tools like ChatGPT, Claude, Grok?
Do you use AI image generators like Midjourney or ImageFX?
What about GrammarlyGo, Notion AI, or anything embedded in apps you already use?
Do you work in a business where AI is very much part of the processes (and do you understand it?)

This could be for work, pleasure, curiosity, research, or even just playing around.

Do you see yourself using AI more in future? Or are you wary of it? Curious? Indifferent?
No right or wrong answers—I'd just love to get a feel for where real people are at with it. It's going to become a much more significant topic over the next 12 months as agentic services mature... happy to explain more if people are interested.

Thanks in advance!

[Taffy on the wing]

I'm exhausted by it already! .......but then i despise this new world of passwords &Apps. Combined with no customer service.
Just last week i sent my sister in London a gift on Amazon uk.....I logged on with my email & password, bought the gift, received a confirmation email & 2 days later an email to track the gift.......it asked for my my email to track it....."sorry we don't recognize that address!"
I tried again with the same result. The AI assistant asked me all kinds of useless questions to try and help.........it's ******* useless as far as i'm concerned!

[Massivefloodlights]

I work at an ‘AI startup’ so I’m familiar with these things, how’s it’s built, and I use AI in my job on top of building it. One of the things I am developing is how to convert what I’ve been developing into the world of agents (AI that does stuff rather than just tell you stuff).

The stuff I build is a mixture of predictive and generative AI - the predictive stuff acting as a foundation for the generative stuff. In my job to be more productive I use things like Gemini and v0 (to create prototypes).

I do a lot of work lecturing and webinar-ing about the future role of AI in my adopted industry (construction) and really AI is a tool which needs to fit a specific job - there’s so much nonsense around that works badly. It’s like a bit of a Wild West at times, a lot of snake oil. And there’s a lot of rubbish AI-generated content (words, images) out there. Fortunately not much on here!

[LincsWoldsClaret]

I use ChatGPT and Claude to check the flow of stuff I’ve written. I find their suggestions for alternative words and phrases become repetitive after time, so I only take their suggestions occasionally. The speed of their processing is amazing. Grammarly is wrong as often as it’s right.

I also use AI for processing photographs - it does the basics far faster than I could do it manually.

I can see my use increasing and I tend to use it as my default search engine. The accuracy still has a long way to go though and its use in customer service environments is abysmal.

[ClaretPete001]

This is an extract from my new book 'Philosophy and being a bit of a Kant'! Out now at all good retailers.

Chapter 3 - The Chimps in the CFS

I think much of the success of AI will depend upon the sophistication of the consumer and how Marketing companies can create the thoughts of individuals as opposed to individuals having their own thoughts.

For example, I find it hard to imagine any kind of AI being able to drum as well as the best drummers in the world because the complexity of it would consume tonnes of terrabytes and cost a fortune. In the end human labour would be quicker and cheaper. The solution to this problem would be to convince the consumer that great drumming is no more worthy than a simple drum machine. And let's be honest, it's already happened and it wasn't hard.

The other big problem for AI is that as it replaces human labour there are fewer consumers to consume the products generated by AI companies. Karl Marx great contribution to human thought was that the value of human labour would become the central source of human conflict. And of course he was right - it will be ..eventually. His silly pamphlet with Engels was just toilet paper in waiting. So, the future consumer will be someone who is a bit dim but relatively manually dextrous: think electrican.

Ergo, the future of AI Marketing companies will be convincing Sparkies that relatively inexpensive products with little intricacy are worth buying. By 2040 the CFS will be half full of trades people shouting 'Who are ya' to people they know full well who they are and the other half will be Marketing Execs studying their behaviour patterns in order to sell them useless junk created by robots for people with the IQ of a chimp.

By that time, the Bob Lord stand will have been shut down because the average citizen will be euthanized by the age of 50. The Jimmy Mac and Longside will be down to the lower tier because the top two tier are full of people whose jobs have been taken by Robots and will be prohibited from urban areas.

[Oshkoshclaret]

First time I used ChatGPT, in December 2022, I was absolutely blown away. It was one of those "I remember exactly where I was when..." moments in life, like learning of Princess Di's death or the 9/11 attacks.

I've had an interest in AI and AGI since the late 90s and the days of the A.L.I.C.E. chatbot. ChatGPT was clearly different gravy. I immediately started using it at work and currently use LLMs (90% ChatGPT, 5% Claude, 5% Grok) fifty to a hundred times per day. It is deeply integrated in every aspect of my life.

I feel this is going to be an utterly revolutionary force in human society. It's going to challenge our very understanding of what it is to be human, as we contend with the reality of an intelligence superior to our own. Anyone in a job that involves knowledge work, of any kind, will be affected. Millions will be out of a job.

Broadly I think it's awesome and incredibly exciting / exhilarating, but we have some challenging times ahead in dealing with it all.

I have been devouring every article, podcast etc I can find on the subject of AI and have found Lex Fridman's podcast to be especially helpful, he's done long form interviews with some of the people at the absolute cutting edge of what has been done, and more importantly what's coming. As a starting point I recommend the one with Dario Amodei at Anthtopic.

In December of 2022 I felt the LLMs were at an 80-90 IQ level. They have steadily progressed and I feel are currently at 140+ in most areas with the exception of maths where it still makes too many mistakes.

There has been a longstanding debate about how the singularity will happen... Fast, almost instantaneous... or slow incremental build up. In my opinion it's clearly the latter and it's clearly happening right now.

[Quicknick]

This is an extract from my new book 'Philosophy and being a bit of a Kant'! Out now at all good retailers.

Chapter 3 - The Chimps in the CFS

I think much of the success of AI will depend upon the sophistication of the consumer and how Marketing companies can create the thoughts of individuals as opposed to individuals having their own thoughts.

For example, I find it hard to imagine any kind of AI being able to drum as well as the best drummers in the world because the complexity of it would consume tonnes of terrabytes and cost a fortune. In the end human labour would be quicker and cheaper. The solution to this problem would be to convince the consumer that great drumming is no more worthy than a simple drum machine. And let's be honest, it's already happened and it wasn't hard.

The other big problem for AI is that as it replaces human labour there are fewer consumers to consume the products generated by AI companies. Karl Marx great contribution to human thought was that the value of human labour would become the central source of human conflict. And of course he was right - it will be ..eventually. His silly pamphlet with Engels was just toilet paper in waiting. So, the future consumer will be someone who is a bit dim but relatively manually dextrous: think electrican.

Ergo, the future of AI Marketing companies will be convincing Sparkies that relatively inexpensive products with little intricacy are worth buying. By 2040 the CFS will be half full of trades people shouting 'Who are ya' to people they know full well who they are and the other half will be Marketing Execs studying their behaviour patterns in order to sell them useless junk created by robots for people with the IQ of a chimp.

By that time, the Bob Lord stand will have been shut down because the average citizen will be euthanized by the age of 50. The Jimmy Mac and Longside will be down to the lower tier because the top two tier are full of people whose jobs have been taken by Robots and will be prohibited from urban areas.

Great title, Pete.

[Quicknick]

If I was teaching in FE/HE again, I wouldn't accept AI-generated essays.

[TheOriginalLongsider]

I take the view that AI can and will enable us to do amazing things. We just need to understand how it works and how to get the best out of it. At work it has speeded up what I do dramatically. For example my employer is upgrading some software from version 6 to 9. That’s quite a big jump, having missed out all the versions in between over 10 years. My job is to plan how we can test that we can unleash this new version without everything breaking. I used ChatGPT to identify all of the differences between the two versions, identify where problems are most likely to occur, what tests I should run, and which teams are the best to do each test. It took about 15 seconds to get results. If I did that without AI it would probably take a couple of weeks of gathering information and asking questions. You can’t just accept whatever it gives you. It’s important to use your own skills to check the output and to tweak it. After that I used Claude to review my test plan - that took about 5 seconds! I can then ask AI to generate whatever style plan or report I require.

It’s worth knowing that the free versions of AI don’t necessarily have up to date information. They take a snapshot of the internet on a particular date. E.g a few months ago I asked it where Burnley would finish in the Championship given their form over the last 5 games and remaining fixtures. The results suggested Kompany was the manager and Leicester were one of our rivals

[AfloatinClaret]

Curious but increasingly wary.
I'm now retired so able - at the moment - to choose how much I interact with AI; or at least I like to think that I'm choosing... I suspect I use AI, or it uses me ten times a day without my even realising it; even predictively typing this?
Wary insofar as I understand AI's advice/opinion/facts are 'learnt' from its interactions with users? In which case the creators/owners, be they government or commercial entities are in the prime position to skew the AI output of advice/opinion and even 'facts' to suit their agendas

[2020visionofleeds]

I use chatGPT to forecast Leeds results, it’s quite accurate.

But it’s not infallible, I asked it to write an account of how Leeds won the European cup and it had the cheek to say it was fictional.

‘Leeds United’s Historic European Cup Triumph

In a stunning chapter of football history, Leeds United lifted the European Cup for the first time in the club’s storied existence. The campaign was a blend of grit, flair, and tactical brilliance, marking a high point in their continental ambitions.

Led by an astute manager and powered by a balanced squad of seasoned internationals and homegrown talent, Leeds navigated a tough knockout path. They overcame European heavyweights with disciplined defending and incisive counter-attacks, turning Elland Road into a fortress. Their midfield, known for its relentless pressing and intelligent distribution, controlled matches, while their striker delivered decisive goals throughout the tournament.

In the final, Leeds faced a formidable opponent in a packed neutral stadium. Undaunted, they executed a flawless game plan. A first-half goal gave them the lead, and a commanding defensive display in the second half sealed a famous 1–0 victory.

The triumph was not just a win for the club, but for a devoted fanbase long hungry for continental glory. Leeds United had finally conquered Europe—and the streets of Leeds erupted in celebration.

Note: This is a fictional account. In reality, Leeds United reached the European Cup final in 1975 but lost 2–0 to Bayern Munich.’

ChatGPT is useless.

[BFCmaj]

I work as an instrumental music teacher. I've used ChatGPT for different things with varying degrees of success. I've found it useful for tasks such as polishing emails, helping to write lyrics for class compositions etc. Which I would class as more creative However, I have found it struggled to do more logical requests such as writing g invoices for private pupils or timetabling lessons in a school. The people who I know who work in creative industries such as music and film are quite worried about the impact that AI might have on their livelihoods. Only this morning I saw an ad for an AI music production app that was pretty scary. Rick Beato has covered a lot about this on his YouTube channel. If you've not seen any of this type if thing how it is being used in music, have a mess around with one of the AI music sites like Udio.com - It's a real eye opener.

[RVclaret]

Couple of interesting AI events this week:

Robot attacks its handler: https://www.dailymail.co.uk/news/articl ... dlers.html

Dead person appears as hologram to deliver victim statement during sentencing for killer: https://www.bbc.com/news/articles/cq808px90wxo

[criminalclaret]

It's very much a tool for the better and it's here to stay, but people have to be realistic about what it's best use cases are at present. I use it as an "aid" in my job but it very much is a "garbage in garbage out" scenario if I don't provide an accurate prompt or context.

I cannot wait to see what it does in the field of medicine and pharmaceuticals.

We've always seen the amazing progress it has made in unveiling protein folding (building blocks of life), can't wait for it usage in other areas of science

Fantastic video btw for anyone interested https://youtu.be/P_fHJIYENdI?si=xBxnLzUX9bCWXz25

[daveisaclaret]

ChatGPT and similar openly available "Generative AI" are basically bad autocomplete algorithms that cost a lot to run and damage the environment.

Scary to think people are turning their livelihoods and personal wellbeing over to this due to what basically amounts to a lack of curiosity.

Although there are doubtless practical applications of actual artificial intellgence being generated which amy in the future be used for good or bad, most of what is publically available today is no more impressive or useful than other tech fads that have raised a lot of money for very unscrupulous people like NFTs.

[Inchy]

I use it all the time, mainly as a replacement for google as it offers direct answers without have to scroll though useless websites.

I find ChatGPT superior to Gemini, which are the two I’ve used.


I’m currently writing my masters dissertation, and although I’m not daft enough to have it formulate written work for me, I’ve found it as a very useful sounding board for ideas.

Asking “is this an appropriate systematic review topic” for example, will break down the question and utilise PICO framework. It will also suggest areas of interest and articles to read.

Universities, and all educational institutions are very wary of its use, but they will have to move with the times. Before the internet everyone spent hours in the library trawling through articles. When internet libraries come about, what took hours/days now took minutes, and also provide the most up to date research globally.

[Leon_C]

Fascinating to see the spectrum of opinion here.

I've literally just come out of an internal presentation to half of our business (a version of a seminar we do to external prospects/clients.)

We explore many of the topics referenced above... I'll offer some jumbled thoughts below:

AI: we ask our guests to be careful not to confuse "ChatGPT" with "AI" broadly - ChatGPT is the single most prevalent branded AI tool, being the first to market in Nov 2022. (I too remember first encountering it, and on my 3/4th prompt casually asking it to debug some troublesome code... I was literally astonished when it fixed it first time.) The field of AI stretches much deeper than ChatGPT use cases.

Prompting: short prompts will deliver vague results in any LLM/chatbot. Good 'prompt engineering' is a life skill...

Context: the fundamental lifeblood of effective response... without relevant business or personal data/information, any AI (LLM) will return generic responses. Re: @2020visionofleeds example - effective forecasting needs substantial historic data as context (if it's not already within the training data). Also - in this example, instructing (via clear prompting) to return only forecasts, and not some creative narrative is important.

Education: really interesting part of the debate - especially around how students are tested and assessed on self-driven work. AI-produced content, in part and full, is now prevalent although detection is possible. However, understanding what we expect students to accomplish will change over time. Being able to test and measure understanding, rather than writing prowess, may become more common.

Reasoning and agency: this is a big part of our education. Whilst millions use ChatGPT as a search engine or a ghostwriter, its emerging capability is problem-solving. We will see much more emphasis on data analysis, resourcing and triage tools that help us with emails, calls, project management. If you can train a human to juggle a handful of spreadsheets, this is fundamentally fair-game for AI whether we embrace it or not.

Superb discussion, I'm equally keen to hear from those who manage to escape it!

[Leon_C]

And as a point of curiosity, one project I'm working on is an image asset generator... here's an example of where it is, at the moment...
Remember that "Today's AI is the worst it will ever be..." - but this takes an existing photo, applies a pre-existing image treatment style and a definitive palette (I created a BFC one in a few seconds) and it knocks out an image.
Now, I don't think any of our team will be thanking AI for what it's done to their faces - but we'll seeing significant improvements to this over the coming months. Also, I'm planning to move my season ticket to that new stand next season...
Anyway, this is a bit of fun for now

[NottsClaret]

Far from an expert so good to get your insight Leon. I think a year or two ago, it seemed like it was taking over the world and our lives were imminently going to be changed forever. Since then, it seems to have calmed down a bit and we're starting to find it's uses and where it's still pretty hopeless, or even damaging.

Also, I reckon we've been using elements of AI for a good while now, maybe they just weren't branded as 'AI'.

[daveisaclaret]

The players don't look human and two of the t-shirts say "We've sterped up" and "Me a Pree"

I don't want to knock your work but what are the human benefits to this over photography and actual art?

[Leon_C]

I don't want to knock your work but what are the human benefits to this over photography and actual art?

Great question, Dave.
This example is an early output from a prototype tool I've built, which assembles an image from (a) a 'real' source image, and a relatively sophisticated treatment prompt. It's built on top of GPT 4o, which is one of the better AI models for image generation.

We can all see the flaws in the actual output, today. But equally, this output took around 60 seconds to generate (a couple of clicks). In theory I could generate thousands a day with the precise same style and format, if there was a commercial reason for me to do so. (If you work in marketing, you can extrapolate use cases here - if you don't work in marketing, then it will seem more tenuous. Also I suspect I'd have some copyright holders coming after me if I carried on, but that's a different matter.)

The point that this makes, is the technology exists to complete a complex task like this in 60 seconds, not so much "do you like this image?" (the image itself is pretty meh, but open to refinement in prompting.) In a few months it take 10 seconds and be much better. In a year, it will be instantaneous, and the facial/detail fidelity will be pretty much perfect. In 2 years, you will likely be generating video in real time, in the same way.

In my presentation, I demo the DALL-E ouputs from 2021 (some of you will know what I mean) compared to today's outputs on the same theme. Mindblowing advances.


What's the benefit over photography and actual art? A debate in the making, yes?
Beauty is in the eye of the beholder. How will businesses find use for this kind of manipulation, and creation of novel content from existing? Advertising, social media, publishing... the clamour for content is massive.

[Leon_C]

Far from an expert so good to get your insight Leon. I think a year or two ago, it seemed like it was taking over the world and our lives were imminently going to be changed forever. Since then, it seems to have calmed down a bit and we're starting to find it's uses and where it's still pretty hopeless, or even damaging.

Also, I reckon we've been using elements of AI for a good while now, maybe they just weren't branded as 'AI'.

Definitely, Notts. We're taking the tools for granted a lot more now. They've become commonplace, and the advances seem more subtle.
If you're familiar with 'reasoning' then this is a useful place to find emerging applications. (Research planning, logical analysis of workloads, that kind of thing.)

Maths was mentioned above. AI isn't particularly adept at maths, as that is a deterministic field (there is a RIGHT answer and a WRONG answer.) Regular code works best for maths - but increasingly, AI models actually write and execute code on the fly to solve problems! Phew.

[CoolClaret]

Crappy, slop 'art' like this has made me (more than ever) realise just how important, and precious human-made art is.

You may think generative (theft) 'AI' for these instances, but I guarantee you that you do not want to live in a world where this stuff is rife.

It's bad enough now with the lack of originality and conformity in popular music/film/art, etc.

[elwaclaret]

If I was teaching in FE/HE again, I wouldn't accept AI-generated essays.

There are tools being employed at most Universities to identify AI generated text. It has become an increasing problem and one of the most challenging areas for markers to cover. I know my lecturers were concerned just how well AI could create essays for classic literature as they were being presented with examples by Edge Hill Universities technical departments in 2023 and it’s only getting more technical. It is a real threat to intellectual copyright and the grading of students (as in identifying what they know as opposed to what they know to ask). I suspect the answer will be that higher degrees will become increasingly obscure, at least in the humanities.

[bpgburn]

And as a point of curiosity, one project I'm working on is an image asset generator... here's an example of where it is, at the moment...
Remember that "Today's AI is the worst it will ever be..." - but this takes an existing photo, applies a pre-existing image treatment style and a definitive palette (I created a BFC one in a few seconds) and it knocks out an image.
Now, I don't think any of our team will be thanking AI for what it's done to their faces - but we'll seeing significant improvements to this over the coming months. Also, I'm planning to move my season ticket to that new stand next season...
Anyway, this is a bit of fun for now

Hells teeth, they look like the cast of The Hills have Eyes..

[Burnleybabe]

My Son who was in The Royal Navy for over 13 years, works in AI in the private sector.
It is mainly defence related that he works on.
Last month he gave a Presentation at NATO Conference in Geneva regarding the safety of in Conflict.
He is only on a NATO sub Committee regarding the safety of AI

[FCBurnley]

I’m old so it’s not for me. It appears to take away responsibility from humans and negates the need for learning ( as does the internet) and will therefore replace a lot of humans in the workplace. Obviously it will have benefits but from what I have seen and read so far the negatives will outweigh them
Good luck with your work

[Leon_C]

Hells teeth, they look like the cast of The Hills have Eyes..

Don't I know it... right now, gpt-4o is pretty weak with faces. We've had some hilarious examples on this testbed.
To be clear - I'm not advocating that the current output standard is nearly good enough for consistent commercial work. The image above is a test, simply to illustrate the point for this thread. But the potential of the tech to create assets (let's not confuse marketing assets with actual art, as they satisfy different societal needs) is significant.

The graphic is something of a red herring, really. AI has much more fundamental applications than knocking out distressed images - although that's where we see and recognise much of its output. As burnleybabe posted, the field of technology extends into defence, cybersecurity, planning.

In fact, 30 minutes ago, I used it to summarise a chain of correspondence from a freeholder and the Land Registry so I could share it with an estate agent.

[Quicknick]

There are tools being employed at most Universities to identify AI generated text. It has become an increasing problem and one of the most challenging areas for markers to cover. I know my lecturers were concerned just how well AI could create essays for classic literature as they were being presented with examples by Edge Hill Universities technical departments in 2023 and it’s only getting more technical. It is a real threat to intellectual copyright and the grading of students (as in identifying what they know as opposed to what they know to ask). I suspect the answer will be that higher degrees will become increasingly obscure, at least in the humanities.

Good post, Elwa. I'm old fashioned. I just regard it as cheating.

[Leon_C]

Good post, Elwa. I'm old fashioned. I just regard it as cheating.

Most major LLMs have a 'deep research' facility. These will, with the click of a button, and perhaps 2-3 clarifying questions, produce PhD standard research documents, in 5-30 minutes. They have access to a wealth of public information, and search the web to complete their research. They don't have access to printed books of course, but increasingly, the majority of useful information is available electronically.

This fundamentally changes what it means to write a paper on a subject, and academia has to work out how it adapts to this. Whilst it's incredibly easy to produce a research paper with this tech, this is a LONG way from researching, and the user gains little or no knowledge of the subject matter.

My view is that 'testing' is likely to require on conversational chatbots (yeah, more AI) which interrogates a student on a topic that they are expected to understand. A reasoning chatbot can steer a line of conversation through a topic, and explore nuance and understanding in ways that a teacher marking a dissertation cannot any more.

[elwaclaret]

Good post, Elwa. I'm old fashioned. I just regard it as cheating.

Me too, sadly it was starting to happen even a couple of years ago during my Masters; fellow students quite openly discussed checking and incorporating AI. Must say I agree with you; I was not interested in getting an award ai hadn’t earned.

[Inchy]

What jobs do you reckon will go with AI?

I know they have used AI to interpret radiological results, and it’s already been shown to be more accurate than a human doctor (I’ve heard this I’ve not seen the study, could be bo11ocks)

When I was growing up everyone was saying get into IT and software engineering. I imagine the vast majority of software engineering jobs will lost to AI.

Taxi drivers could go. They are already using self driving taxis in America.

A lot media jobs could go. Editors, writers, etc.

[Inchy]

Good post, Elwa. I'm old fashioned. I just regard it as cheating.

I imagine people used to say that about the abacus… and then the calculator.


AI is not leaving. Universities have to catch up.

Personally I wouldn’t use it for anything but a sounding board, which it’s great for. I actually want to learn as it helps my patients, and AI cannot do physical assessments…….until they become robotic and them I’m knackered

[Inchy]

Most major LLMs have a 'deep research' facility. These will, with the click of a button, and perhaps 2-3 clarifying questions, produce PhD standard research documents, in 5-30 minutes. They have access to a wealth of public information, and search the web to complete their research. They don't have access to printed books of course, but increasingly, the majority of useful information is available electronically.

This fundamentally changes what it means to write a paper on a subject, and academia has to work out how it adapts to this. Whilst it's incredibly easy to produce a research paper with this tech, this is a LONG way from researching, and the user gains little or no knowledge of the subject matter.

My view is that 'testing' is likely to require on conversational chatbots (yeah, more AI) which interrogates a student on a topic that they are expected to understand. A reasoning chatbot can steer a line of conversation through a topic, and explore nuance and understanding in ways that a teacher marking a dissertation cannot any more.

I’m doing my masters in Advanced Clinical Practice. Assessment is a mixture of clinical assessment, essays, and exams.

I’ve done other higher educational module in the past as well. I always find I learn more from exams than essays because you don’t know what’s in the exam so you have to learn everything.

I have to write a systematic review for my dissertation (no doubt AI could do this in minutes). I’ve also written systematic reviews and discursive reviews in my undergrad. You can get away with passing essays with minimal learning, in my experience. But exams you actually have to learn and memorise. I think it will go that way in many areas. Not easy for some subjects though

[JohnMac]

I delved into Chat GPT a couple of weeks back and asked it to produce a Birthday card for a Grandson. I supplied a photo with a suggestion and it did a really good job.

I uploaded a photo of me with a suggestion and was impressed with how intuitive it can be and handle amendments.

That will be the limit of my involvement, just messing about. As a Club photographer with an interest in Competitions it is a definite no no using AI but an AI image against my finished one may be an interesting comparison for learning purposes.

[The Shire Claret]

AI has resulted in thousands of businesses most likely spending way too much money to bulk emailing business's which is become increasingly tedious to unsubscribe from them all

Waste of money on that front

Other then that - it can be quite useful

[brexit]

a paper I co-authored
https://www.lesswrong.com/posts/TpSFoqo ... oks-like-1

[Poulton-le-Claret]

Quite enjoyed this YouTube channel I stumbled on the other day. Uses AI to make previews for movies/series in different styles like 1950s panavision. Not perfect by any stretch, but still impressive.

https://youtube.com/@abandonedfilms?si=fNiyKiF-FO_f5nqQ

[LoveCurryPies]

It’s going to increasingly more sophisticated and difficult to know what’s real and what isn’t.

[CoolClaret]

I’m doing my masters in Advanced Clinical Practice. Assessment is a mixture of clinical assessment, essays, and exams.

I’ve done other higher educational module in the past as well. I always find I learn more from exams than essays because you don’t know what’s in the exam so you have to learn everything.

I have to write a systematic review for my dissertation (no doubt AI could do this in minutes). I’ve also written systematic reviews and discursive reviews in my undergrad. You can get away with passing essays with minimal learning, in my experience. But exams you actually have to learn and memorise. I think it will go that way in many areas. Not easy for some subjects though

I'm doing one in an combined Engineering field (Electrical/Mechanical + some other elements). I can categorically state that an AI cannot produce MSc standard (let alone PhD standard) research docs in 5-30 minutes.

[Leon_C]

I can categorically state that an AI cannot produce MSc standard (let alone PhD standard) research docs in 5-30 minutes.

Yeah - I somewhat misspoke.
OpenAI claim that their TOP end agents can operate at PhD level.. that isn’t the capability of the £20 per month flavour just yet.

[Dressinggown]

It's strange what AI can do to the photography, art and graphic design world.



Here are my elderly relatives celebrating our promotion in their care home.

[Ric_C]

And as a point of curiosity, one project I'm working on is an image asset generator... here's an example of where it is, at the moment...
Remember that "Today's AI is the worst it will ever be..." - but this takes an existing photo, applies a pre-existing image treatment style and a definitive palette (I created a BFC one in a few seconds) and it knocks out an image.
Now, I don't think any of our team will be thanking AI for what it's done to their faces - but we'll seeing significant improvements to this over the coming months. Also, I'm planning to move my season ticket to that new stand next season...
Anyway, this is a bit of fun for now

Remember that "Today's AI is the worst it will ever be..."

Is that you VK?

[Sutton-Claret]

I feel i need to learn it to keep up with Techno - But what is the best to learn, I've used Chat Gbt but every week a new one comes out stating to be the best? Give me one thats going to be around and relevant for a few years and I'll find a youtube course

[Leon_C]

I feel i need to learn it to keep up with Techno - But what is the best to learn, I've used Chat Gbt but every week a new one comes out stating to be the best? Give me one thats going to be around and relevant for a few years and I'll find a youtube course

ChatGPT isn't going anywhere. It's as good a place to start as any of the others (it's the one i use most often.)

How to get started?

First - some experimentation for daily life information - curiosity.
1) give me 5 interesting dinner recipes that are easy to prepare. my favourite 5 ingredients/flavours are [1,2,3,4,5]
2) produce an itinerary for a 3 day trip to (York, Edinburgh, Paris, Buenos Aires.) The kind of things I like to do are [1,2,3]
3) give me some ideas for interesting photographic projects that I can do with my phone. I live in Burnley, and I enjoy taking photos of [flowers, people, buses, angry wasps]

These kind of things give you a sense of the possibilities. Crucially - the context, that is, the background for the answers above, is the difference between generic responses, and specific, useful responses. ChatGPT - as the name suggests - is a chatbot. It has a degree of memory for your conversation, so you can drill down into the answers more. Reject some of the recipes above, for instance, and it will suggest more. Tell it that you only have 2 days in York, and you have a dinner reservation at 1800, and it will adjust the itinerary around that.

You can get the ChatGPT mobile app, which means that you have some interesting use cases, if this kind of thing interests you.

Example: I was in France in January. Having dinner with some friends, and whilst I speak decent French, and the menu was translated - this doesn't really give you a true sense of the more local dishes. I uploaded photos of the menu, and asked it what it would recommend. Within a few seconds, I had a list of accurate interpretations for the more 'sophisticated' dishes, which made them more appealing - and helped me to choose.

Take a photo of a plant, and it will try to identify it (and even indicate any potential health issues with it.) There are other apps that do this, of course.

[nig1954]

I’m an old codger but I’ve already been advised that AI agents will become a way of life within the next 5 years and that I need to adapt or get left behind

[Sutton-Claret]

ChatGPT isn't going anywhere. It's as good a place to start as any of the others (it's the one i use most often.)

How to get started?

First - some experimentation for daily life information - curiosity.
1) give me 5 interesting dinner recipes that are easy to prepare. my favourite 5 ingredients/flavours are [1,2,3,4,5]
2) produce an itinerary for a 3 day trip to (York, Edinburgh, Paris, Buenos Aires.) The kind of things I like to do are [1,2,3]
3) give me some ideas for interesting photographic projects that I can do with my phone. I live in Burnley, and I enjoy taking photos of [flowers, people, buses, angry wasps]

These kind of things give you a sense of the possibilities. Crucially - the context, that is, the background for the answers above, is the difference between generic responses, and specific, useful responses. ChatGPT - as the name suggests - is a chatbot. It has a degree of memory for your conversation, so you can drill down into the answers more. Reject some of the recipes above, for instance, and it will suggest more. Tell it that you only have 2 days in York, and you have a dinner reservation at 1800, and it will adjust the itinerary around that.

You can get the ChatGPT mobile app, which means that you have some interesting use cases, if this kind of thing interests you.

Example: I was in France in January. Having dinner with some friends, and whilst I speak decent French, and the menu was translated - this doesn't really give you a true sense of the more local dishes. I uploaded photos of the menu, and asked it what it would recommend. Within a few seconds, I had a list of accurate interpretations for the more 'sophisticated' dishes, which made them more appealing - and helped me to choose.

Take a photo of a plant, and it will try to identify it (and even indicate any potential health issues with it.) There are other apps that do this, of course.

Thanks Leon - all very helpful. I will give it another chance over the weekend.

[CoolClaret]

ChatGPT isn't going anywhere. It's as good a place to start as any of the others (it's the one i use most often.)

How to get started?

First - some experimentation for daily life information - curiosity.
1) give me 5 interesting dinner recipes that are easy to prepare. my favourite 5 ingredients/flavours are [1,2,3,4,5]
2) produce an itinerary for a 3 day trip to (York, Edinburgh, Paris, Buenos Aires.) The kind of things I like to do are [1,2,3]
3) give me some ideas for interesting photographic projects that I can do with my phone. I live in Burnley, and I enjoy taking photos of [flowers, people, buses, angry wasps]

These kind of things give you a sense of the possibilities. Crucially - the context, that is, the background for the answers above, is the difference between generic responses, and specific, useful responses. ChatGPT - as the name suggests - is a chatbot. It has a degree of memory for your conversation, so you can drill down into the answers more. Reject some of the recipes above, for instance, and it will suggest more. Tell it that you only have 2 days in York, and you have a dinner reservation at 1800, and it will adjust the itinerary around that.

You can get the ChatGPT mobile app, which means that you have some interesting use cases, if this kind of thing interests you.

Example: I was in France in January. Having dinner with some friends, and whilst I speak decent French, and the menu was translated - this doesn't really give you a true sense of the more local dishes. I uploaded photos of the menu, and asked it what it would recommend. Within a few seconds, I had a list of accurate interpretations for the more 'sophisticated' dishes, which made them more appealing - and helped me to choose.

Take a photo of a plant, and it will try to identify it (and even indicate any potential health issues with it.) There are other apps that do this, of course.

Are you not worried that you're automating your life/brain away?

Seems like a way of life that I'd genuinely hate to live by.

How long until advertising influences AI so what you may think is objective choices to fit your requirements is just pulling you into whichever establishment/shop has paid the most money?

No offence, but your recommendations there seem really, really sad.

One of the most enjoyable and imo, important parts of life is 'adventure' - it's not much of an adventure if you're marching to the beat of a LLM, is it?

[Leon_C]

No offence, but your recommendations there seem really, really sad.

I'm going to refrain from answering every point - but when somebody at points at a topic that many people find interesting, and declare that they're "really, really sad" then this says more about them. You do you.

[CoolClaret]

I'm going to refrain from answering every point - but when somebody at points at a topic that many people find interesting, and declare that they're "really, really sad" then this says more about them. You do you.

You know what is interesting? Being human and having real human thoughts/feelings/emotions.

My PoV there was qualified with salient points/questions that I posed to you but you chose to ignore and attempted to deflect.

If you can't see the problem with automating your entire decision process away for things as simple as leisure time activities, then I don't know what to tell you.

God speed.