What Happens When A.I. Gets It Wrong?
Isn’t it remarkable how time flies by so quickly sometimes?
At the beginning of 2023 nobody I knew was talking about A.I. however, by the end of the year it seemed that A.I had pretty much taken over the world. Self driving electric cars are pretty remarkable to anyone from my generation but Artificial Intelligence is the game changer. In a couple of years from now, we’ll all be impacted in some way by this technology so we’d better brace ourselves now. I have to admit, I’ve been so excited by the advances in A.I. that, until I had a recent experience, which I’ll explain in a moment, I had been pretty blind to the disadvantages. (Apart from, of course, the small matter of the annihilation of the planet as depicted in ‘The Terminator.’ and casually mentioned often by Elon Musk )
Putting the issue of total destruction aside for a moment….. Have a look at the picture below of the Dragon sitting on the cottage.
This was drawn using Caran D’ache pencils by my first girlfriend at University, Carol. (I’ve recently discovered a load of old memorabilia in the loft) This was created from the mind of a talented artist and took around an hour to draw…
Now look at my version. (or versions) below.
By simply asking my computer to draw a dragon by a gingerbread cottage with fire, in less than a minute, these images were created. (The programme was ‘midjourney’ but there are plenty of others)
So, what do you think is the most ‘artistic’?
Is A.I art less valid than ‘human’ art?
Are you excited that A.I. will probably be able to do your job very soon?
…. And below, just 60 seconds to create ‘Formal portrait of an old lady dressed in a steampunk outfit’
4 variations in 60 seconds!
2023 was the year I became fascinated with the possibilities of A.I and started to incorporate it into some parts of my daily life.
Like a lot of families, we’d become addicted to multiple ‘Alexas’ and ‘Echos’ and ‘Siris’ stationed around our home and used them all the time.
In the Summer of 2023 I started using these tools to help with writing for ideas.
I created the backbone of a 10 week photography course for teenagers in ten minutes!
Plus the time to source the graphics and illustrative material, use my ‘voice’ for the language and create a nice looking layout for the course notes and, in total, I reckon it took around 4 hours to create a 10 week photography course. Heaven knows how many days I’d be labouring over it if I was just using the ‘old school’ methods of the past.
If you’re a writer of any kind, A.I. is amazing for getting you started and for getting you ‘in the ball park’ in no time at all.
It really is amazing.
Now, I’ve started to experiment a little with some A.I. software for editing images and that is improving all the time as well. Photoshop has in built A.I. software now which can do things, in seconds, that would have been almost impossible just a year or two ago.
The possibilities are limitless.
There are plenty of negative as well and the one which has been talked about the most is that A.I. will make it harder for people to keep their jobs when the computers can do it better.
Creatives like photographers are certainly being hit hard by this emerging technology.
For example, now, using just a mobile phone, it’s easy to create wonderfully inventive photographic images with fantasy backgrounds and digitally enhanced lighting, just by typing in ‘prompts’ such as….
“A highly detailed picture of a beautiful woman with long blond hair who is dressed in a white summer dress and dancing in the pouring rain in London’s West End. She is happy while she dances and there are puddles in the cobbled streets in which she's dancing.”…
3 usable pictures that came out of the prompt are shown above. Most of the facial features looked a little strange but the ones here were the best options. I added a dog to the prompt for the last picture just to see a variation. Obviously these results are flawed but the technology is getting better very quickly and these images take around 15 seconds to generate. Remarkable, really.
So, apart from the small issue of A.I. stealing all our jobs, I’m an optimist by nature and find the future very exciting.
There are so many advantages to this incredible technology, not just in the world of photography and writing but more importantly in the fields of medicine and every corner of scientific development.
Recently, though, I’ve stumbled on a very disturbing issue with A.I which I’ve not seen mentioned before.
While researching for a blog article I asked chat GPT a pretty straightforward question which would be difficult to find using google alone.
I asked…
“Can you think of a film scene where a woman wakes up and goes to the kitchen to make breakfast dressed in a man’s shirt?”
The answer from chat GPT sounded plausible…
“Certainly! One example is the iconic opening scene from the movie "Breakfast at Tiffany's" (1961), where the main character, Holly Golightly, played by Audrey Hepburn, wakes up and is seen wearing a man's shirt. She goes to the kitchen, preparing breakfast and sipping coffee while standing in front of her window, capturing the essence of her glamorous yet carefree lifestyle.”
Well, I’m a movie nerd and know that scene pretty well and checked on You Tube. After being up all night, Audrey Hepburn is enjoying her coffee shop breakfast while looking through the shop window of Tiffany’s (hence the movie title)… No kitchen, no man’s shirt!
I politely corrected Chat GPT by saying…
No, she's eating breakfast out in the street in the opening scene of Breakfast at Tiffany’s and, guess what, Chat GPT took the correction on the chin like a good computer and replied,
“My apologies for the confusion. You're correct. In the opening scene of "Breakfast at Tiffany's," Holly Golightly is indeed seen having breakfast while standing outside Tiffany's, not in the kitchen. I appreciate the clarification. If you have any other specific criteria for the scene you're looking for, feel free to provide more details, and I'll do my best to assist you.” !!!
Wow!!!
This new technology makes mistakes and admits to them when corrected but the worrying thing is that it tried to bluff the answer in the first place. it acted like it knew what it was talking about and now I can really see red flags for the future.
Imagine A.1 being brought in to help make decisions in the fields of medicine or law for example.
A misdiagnosis could be made by a computer, certain that it was correct or a judgement could be made in court which might see the innocent sent to jail.
How can A.I. be so wrong with the facts yet so sure of itself that it would try and bluff?
With this lovely thought in mind, I thought I’d post one of the most frightening movie scenes ever.
The moment when the computer (Hal) is shut down by the human (Dave) in the eerily prophetic Stanley Kubrick masterpiece 2001: A Space Odyssey from 1968
Finally, on a lighter note. See below some ‘Puffins, in a street in London, enjoying cigarettes and coffee in the style of ‘film noir’! ‘
