Chat GPT

The world of AI is at a very exciting stage. I'd love to know if City plan to make use of AI or ar doing so already. But in all fields, music, art, literary, and more its going to have a massive impact.

I asked chatgpt to write me a short war story, each time it was different, I asked it for happy, sad endings, ironic endings... Very clever at this early stage.

This was interesting too:

 
As an AI language model, I cannot predict future events. However, it is important to note that Man City has consistently been one of the top-performing teams in the Premier League and has a talented squad with skilled players. With the right management and strategies in place, they have the potential to win the Premier League in 2023. But ultimately, the outcome of the season will depend on a variety of factors such as player injuries, other teams' performances, and unexpected events that may impact the league standings.

Answer to will Man City win the EPL 2023
 
Write a song about diarrhoea in the style of Metallica

Verse 1:
Pain in my gut, can't take it no more
Stomach churns, it's like a war
Something's brewing deep inside
Can't escape, can't hide

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Verse 2:
Metallica's playing loud and clear
But my bowels are full of fear
Sweat drips down, it's not from the show
Wishing this moment would go

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Bridge:
It's like a battle in my gut
Can't escape, can't shut
The explosive force that's unleashed
Leaves me weak, feeling deceased

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Outro:
Metallica's riffs fade away
As I wash the stains of dismay
The pain subsides, but the memory remains
Of that time I had diarrhoea pains.
 
Write a song about diarrhoea in the style of Metallica

Verse 1:
Pain in my gut, can't take it no more
Stomach churns, it's like a war
Something's brewing deep inside
Can't escape, can't hide

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Verse 2:
Metallica's playing loud and clear
But my bowels are full of fear
Sweat drips down, it's not from the show
Wishing this moment would go

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Bridge:
It's like a battle in my gut
Can't escape, can't shut
The explosive force that's unleashed
Leaves me weak, feeling deceased

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Outro:
Metallica's riffs fade away
As I wash the stains of dismay
The pain subsides, but the memory remains
Of that time I had diarrhoea pains.
Poetry in motion
 
Write a song about diarrhoea in the style of Metallica

Verse 1:
Pain in my gut, can't take it no more
Stomach churns, it's like a war
Something's brewing deep inside
Can't escape, can't hide

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Verse 2:
Metallica's playing loud and clear
But my bowels are full of fear
Sweat drips down, it's not from the show
Wishing this moment would go

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Bridge:
It's like a battle in my gut
Can't escape, can't shut
The explosive force that's unleashed
Leaves me weak, feeling deceased

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Outro:
Metallica's riffs fade away
As I wash the stains of dismay
The pain subsides, but the memory remains
Of that time I had diarrhoea pains.
Better than they ever managed :)
 
Write a song about diarrhoea in the style of Metallica

Verse 1:
Pain in my gut, can't take it no more
Stomach churns, it's like a war
Something's brewing deep inside
Can't escape, can't hide

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Verse 2:
Metallica's playing loud and clear
But my bowels are full of fear
Sweat drips down, it's not from the show
Wishing this moment would go

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Bridge:
It's like a battle in my gut
Can't escape, can't shut
The explosive force that's unleashed
Leaves me weak, feeling deceased

Chorus:
Diarrhoea, oh no
It's taking control
Running to the bathroom, can't hold
Diarrhoea, oh no

Outro:
Metallica's riffs fade away
As I wash the stains of dismay
The pain subsides, but the memory remains
Of that time I had diarrhoea pains.
Write a song about ChatGPT writing a song about diarrhoea in the style of Metallica

Intro:
I'm ChatGPT, and I'm here to say
I'm gonna write a song in a metal way
It's about a topic that's not so pretty
But it affects us all, it's not so witty

Chorus:
Diarrhoea, it's a pain in the gut
Runs right through you, it's hard to cut
But with some rest and fluids to drink
You'll be back to normal in a blink

Verse 2:
It's not a topic that's often discussed
But it's a fact of life, it's a must
It can be caused by many things
From viruses to spicy chicken wings

Chorus:
Diarrhoea, it's a pain in the gut
Runs right through you, it's hard to cut
But with some rest and fluids to drink
You'll be back to normal in a blink

Bridge:
You gotta stay hydrated, don't let it get you down
Drink lots of water, it'll turn your life around
It's not a laughing matter, it can be quite severe
But with some TLC, you'll soon be in the clear

Chorus:
Diarrhoea, it's a pain in the gut
Runs right through you, it's hard to cut
But with some rest and fluids to drink
You'll be back to normal in a blink

Outro:
So now you know, it's not so bad
Just take care of yourself, and you'll be glad
Thanks for listening, I hope you had fun
Stay healthy and strong, rock on and on!
 
Started playing around with Bard and suspect will be using some sort of version in work soon
 
Has any in plus got access to the plugins? It says they are available but nothing there when logged in under settings
 
used chat-gpt today to write some technical specification clauses. It came out with very generic stuff and was useless in terms of writing the clauses, but what it did do was provide a decent tick list of things to include. Useful in that regard but can't replace me yet
 
And just keeping the thread alive, the ability of current AI, to create images like below, in seconds, with just a few prompts, is truly staggering. Bear in mind, this girl does not exist. Currently we are limited to photos, but when it can produce video of this quality - which is right around the corner, btw - then god knows where that will lead.

midjourney-v5-drafts-Artboard-4-1536x768.jpg
 
And just keeping the thread alive, the ability of current AI, to create images like below, in seconds, with just a few prompts, is truly staggering. Bear in mind, this girl does not exist. Currently we are limited to photos, but when it can produce video of this quality - which is right around the corner, btw - then god knows where that will lead.

midjourney-v5-drafts-Artboard-4-1536x768.jpg
I just tried to use generative fill, which is the new AI feature on Photoshop and it was wank. Tried to put a hat on my girlfriend and it could add a hat, but it couldn't really make it look like she was wearing it. It didn't seem to have any clue about scale. Seemed to get the lighting right though. Tried to change her from smiling to frowning, and it just pasted a new face over the top of hers. It seems like they're able to create impressive images from scratch, but they're awful at identifying what's already in an image and modifying it, let alone making it look like it belongs in the scene. As with everything, a lot of hype on Youtube but when you get your hands on it, it's not all that.

The best feature is actually the one we've had for a few years now, which is Select Subject. It usually does a pretty good job of identifying the subject of the image and then you can edit it separately from the rest of the image. I use that one all the time.
 
I just tried to use generative fill, which is the new AI feature on Photoshop and it was wank. Tried to put a hat on my girlfriend and it could add a hat, but it couldn't really make it look like she was wearing it. It didn't seem to have any clue about scale. Seemed to get the lighting right though. Tried to change her from smiling to frowning, and it just pasted a new face over the top of hers. It seems like they're able to create impressive images from scratch, but they're awful at identifying what's already in an image and modifying it, let alone making it look like it belongs in the scene. As with everything, a lot of hype on Youtube but when you get your hands on it, it's not all that.

The best feature is actually the one we've had for a few years now, which is Select Subject. It usually does a pretty good job of identifying the subject of the image and then you can edit it separately from the rest of the image. I use that one all the time.
It's a huge hit and miss. It promises so much but ends up consuming a lot of time. I'm not at all convinced it's at all ahead of (very easy to use and totally free) software like NMKD and an inpainting model from Civitai.

The hit and miss thing is everywhere with current AI. The cynic in me thinks this is perfect for hooking people. You do keep going back, even after hours of wasted work.

Technologically, its due to the current generative AI technologies epitomising 'one shot' generative AI... which is just one application of the underlying technology... it happens to be the easiest to release as a product.

Sometimes this one shot stuff infers what you have in mind perfectly. Those moments are still pretty astounding. But when it's off, getting it back on track involves an awful lot of time consuming fiddly work.

These systems need to integrate feedback and control from the user before they deliver, in order to reach their potential. Integrating that is a big challenge. A1111 Stable Diffusion Web UI with Controlnet has come the closest by far - at least for image work. I couldn't think it's much use without GIMP or Photoshop in the background tho.

GPT and the like were trained on text and still images. There is some hope that training them on moving images will help the models "relate" to us and the world we live in.... relate is the wrong word. It's to encode fundamental things in the model about aspects of reality like cause and effect, physics... making it more useful, reliable, grounded. I can see it, to some extent. Personally I think in order to be useful, they will need to do a massive, massive amount of training on our phone sensors and computer inputs / outputs. That would focus it on the work it can achieve for us.

There's definitely a fine balance between hype and reality with all of this. But there was with the internet, twenty five years ago. The danger is that we get systems that keep us more hooked than ever, but don't deliver much in terms of productivity, efficiency, human happiness, quality of life.

We've seen this sort of thing all through the information revolution. Every advance creates more work and confusion. You can't take the human element out of the system, as there is no system that wasn't drawn to meet the demands of humans, who are always selfish, short sighted, politically and economically motivated.

The danger with AI's is that they reflect that more accurately than ever before - investment, in economic terms, works on the promise of tomorrow. Everyone knows there's a problem with that. Deliver, and the investor 'leaves'. So it pays to keep the investor hooked forever, and that does not neccessarily mean delivering anything worthwhile.


On a societal scale this means we will quite likely end up deeper than ever in a never ending game of chasing our tail, thinking we're getting somewhere, whilst dying younger, in a less pleasant environment.... etc.

Sometimes all that is required to keep us from thinking clearly, is the sensation of progress.
 
TBH Jim, that was the last epoch of the technological revolution - Big Data.

It happened with the internet and mobiles, especially when they came together.

Also think... track and trace, chip and pin, bluetooth, wifi, CCTV. Clubcard. Any and all websites who take your email address. The move to Gmail and Hotmail, rather than an account with your ISP. Captchas. 3G, 4G. Echo and now smart TVs.

Regarding the mobiles and your web browser, there is constant tracking which you are never aware of. It quickly reached a level where there was too much information being sent to too many places for anyone to actually think about. The good news is the information is pretty boring, no one really likes looking through it on an individual level. Rather than control all aspect of YOUR thoughts individually, they use it en masse come up with models and algorithms that nudge you in a certain direction, keep you spending, keep you viewing.
 
It's a huge hit and miss. It promises so much but ends up consuming a lot of time. I'm not at all convinced it's at all ahead of (very easy to use and totally free) software like NMKD and an inpainting model from Civitai.

The hit and miss thing is everywhere with current AI. The cynic in me thinks this is perfect for hooking people. You do keep going back, even after hours of wasted work.

Technologically, its due to the current generative AI technologies epitomising 'one shot' generative AI... which is just one application of the underlying technology... it happens to be the easiest to release as a product.

Sometimes this one shot stuff infers what you have in mind perfectly. Those moments are still pretty astounding. But when it's off, getting it back on track involves an awful lot of time consuming fiddly work.

These systems need to integrate feedback and control from the user before they deliver, in order to reach their potential. Integrating that is a big challenge. A1111 Stable Diffusion Web UI with Controlnet has come the closest by far - at least for image work. I couldn't think it's much use without GIMP or Photoshop in the background tho.

GPT and the like were trained on text and still images. There is some hope that training them on moving images will help the models "relate" to us and the world we live in.... relate is the wrong word. It's to encode fundamental things in the model about aspects of reality like cause and effect, physics... making it more useful, reliable, grounded. I can see it, to some extent. Personally I think in order to be useful, they will need to do a massive, massive amount of training on our phone sensors and computer inputs / outputs. That would focus it on the work it can achieve for us.

There's definitely a fine balance between hype and reality with all of this. But there was with the internet, twenty five years ago. The danger is that we get systems that keep us more hooked than ever, but don't deliver much in terms of productivity, efficiency, human happiness, quality of life.

We've seen this sort of thing all through the information revolution. Every advance creates more work and confusion. You can't take the human element out of the system, as there is no system that wasn't drawn to meet the demands of humans, who are always selfish, short sighted, politically and economically motivated.

The danger with AI's is that they reflect that more accurately than ever before - investment, in economic terms, works on the promise of tomorrow. Everyone knows there's a problem with that. Deliver, and the investor 'leaves'. So it pays to keep the investor hooked forever, and that does not neccessarily mean delivering anything worthwhile.


On a societal scale this means we will quite likely end up deeper than ever in a never ending game of chasing our tail, thinking we're getting somewhere, whilst dying younger, in a less pleasant environment.... etc.

Sometimes all that is required to keep us from thinking clearly, is the sensation of progress.
There's some good points in there and I agree with some and others less so. But the one thing I would say is that it is impossible to draw conclusions based on previous parallels, because there have never been any. AI in it's current state of play (and that which really is just around the corner) is literally the biggest thing to ever happen in the history of mankind. It's that profound. The internet, electricity, the car etc - these are all minor in comparison to the monunmental significance AI is about to have on the development - or fate - of our species.

People seem to think that generative AI is pretty dumb and it's basically just predicting what to say next. In a way, it is. But if you want to predict what someone is going to say next - beyond a word or two - you have to UNDERSTAND what it is they are saying. AI can do this now. It really, truly, understands. Whether current models are sentient yet is a needless distraction, not least because no-one can really define what sentience is. But do these models "think"? Absolutely they do.

They now have the ability to take in unstructured data - text, video, sound etc and by looking for patterns, figure our what it means. Remarkably, they are already way better at learning than humans are. The largest language models have something like 100x less neural connections than the human brain and yet are already 100x more knowledgeable in terms of the amount of information they can draw upon. And here's a critical thing - they can pass information from one instance to another at unimaginable speed - many gigabytes per second - whereas humans can manage a few words a minute.

It is rapidly becoming beyond debate that within the next few years we will have AI's more intelligent than humans and a few years beyond that, immeasurably more intelligent. And in 2 or 3 hundred years time, it is quite possible that outsiders would look at planet Earth and see that all intelligent "life" on the planet is digital and that human intelligence was a mere fleeting novelty. Is that likely? Who knows. No-one knows. We are walking inexorably into a dense fog with no idea what lies ahead.
 
TBH Jim, that was the last epoch of the technological revolution - Big Data.

It happened with the internet and mobiles, especially when they came together.

Also think... track and trace, chip and pin, bluetooth, wifi, CCTV. Clubcard. Any and all websites who take your email address. The move to Gmail and Hotmail, rather than an account with your ISP. Captchas. 3G, 4G. Echo and now smart TVs.

Regarding the mobiles and your web browser, there is constant tracking which you are never aware of. It quickly reached a level where there was too much information being sent to too many places for anyone to actually think about. The good news is the information is pretty boring, no one really likes looking through it on an individual level. Rather than control all aspect of YOUR thoughts individually, they use it en masse come up with models and algorithms that nudge you in a certain direction, keep you spending, keep you viewing.
I was aware that "big brother" can track you via your mobile. Even (supposedly) when the phone is switched off.
 
As someone who teaches masters students, it's a complete nightmare for us. We have a lot of Chinese students and it's quite remarkable how many of them have suddenly developed an incredible English vocabulary (one essay I marked last week had the word 'panacea' in it and the lad struggles with basic verbal English). The similarity checks can't pick it up and we're just having to make do. We're trying to think of ways to mitigate against it, getting students to critique a chatGPT created essay for example, but it really has thrown a lot of things on their head for us.
My son-in-law’s a teacher and he called out a couple of students who suddenly turbo-boosted their language skills. However, as you suggest, this is likely to be an increasingly futile battle.

My son’s a bit of an author and poet and he’s amazed at even the present capability of this app.
 
I've been using since it was made public, I find it very useful as a programmer but the improvements it's made in recent months has me worried to the point I'm looking a career options. More so Auto gpt which I played around with and it programed a WordPress plugin with very little promts, scary but impressive stuff.
 
It cant think in the same way a human thinks, it isnt thinking at all. Its being trained on obscenely large amounts of data that we cant even fathom. The problem with visualising how large language models work is that people dont understand how computers work, they can do billions of operations a second.
There is enormous effort going into getting this result, and that is through the use of massive data and massive computing power.
 
Last edited:

Don't have an account? Register now and see fewer ads!

SIGN UP
Back
Top