Imagine… Artificial Intelligence How bad could it be? Story by Eleonora de Gray, Editor-in-Chief of RUNWAY MAGAZINE. Photos: Concept Runway Magazine, generated by AL, retouched and redesigned by Runway Magazine.
Artificial Intelligence and fear
Artificial Intelligence… How scary could it be? Artificial Intelligence is the simulation of human intelligence processes by computer systems. AI involves the development of algorithms and computer programs that can perform tasks that typically require human intelligence, such as visual perception, speech recognition, and language translation. The goal of AI is to create intelligent machines that can work and think like humans.
Well… today AI is not even close to being a simulation of human intelligence. It is very advanced programming, which is still not capable to analyze, not even capable to compare the facts or images, or provide correct information about the events. And certainly AI is not capable to analyze or create.
Today AI is a tool, an advanced tool, a software for creators, to help them to create faster. But it doesn’t mean fast enough to call it “fast fashion” or “fast design”.
What is the most important in the creative process? Is the idea, the concept.
So when I hear whispers of scary institutions I’m surprised. Harvard Business School recently published an article about generative AI and Intellectual Property problems. May… or may not have? That is a question.
There are two of the most advanced AI robots which blew up minds: Image Generator MidJourney and Text Generator ChatGPT. Well there’s also Bing Chat (in search engine) and Google text corrector (in Gmail). All of them are based on text generation technologies. Yes, all of them, including MidJourney.
The article published on Harvard Business School website started with the description of an art exhibition organized by The Museum of Modern Art in New York, who hosted an AI installation generated from the museum’s own collection, etc. Speed and quality generated visuals etc. Then in this article we read about “striking poems and jokes” generated by AI.
Did you actually try to AI robot ChantGPT or BingChat to tell you a joke? I tried. Believe me… it’s not funny. Even worse when you ask to generate the joke. This is simply dull.
AI robot is a generator, not a creator. So no art, no creative thinking, no jokes are possible for AI robot.
So the fear is coming from the fact that the robot takes human creations, images and texts, and generates its own. Is it really?
So then Harvard researcher immediately without research on the subject raised the legal question: if an AI robot took visuals from the authors and generated its own to who owned the ownership of these visuals. And what about the copyright?
It goes even further. French institutions, schools and federations are panicking, screaming in fear: “AI risks bringing out a “fast creation”, like “fast fashion” in fashion”, said Le Monde. Does it really?
There is also some reflective analyses. Les Echos: “More than half of French people see artificial intelligence as a threat. This is all the more surprising since they are the ones who buy the most consumer goods equipped with AI… Before sinking into a replica of Terminator and Skynet, it is important that everyone is aware that AI will truly revolutionize our daily lives. Our way of moving, consuming and even thinking. New professions will appear between the philosopher and the developer, and it is precisely at this time that the notion of ethics will be essential.”
But between the fair, panic, legal questions and analyses, has anyone really tried to read the manual and tried how it works? It really helps to understand AI.
AI and what it really is today – Knowledge is the antidote to fear
“Fear comes from the lack of knowledge and a state of ignorance. The best remedy for fear is to gain knowledge.” ― Debasish Mridha.
How true. So let’s get this knowledge from reading the manual and having own personal experience with AI.
I’m not going to go into the explanation about human culture, language, and where our logic came from. Yes, it is language. So today all known robots are based on operating with the text. What they really can do – it’s to pull our knowledge database for you. Maybe compare it a little. And That’s it.
What do you think it really is? Yes, this AI robot can pull you knowledge database, it also has knowledge about different styles, and orthography. But only you can find a solution to a problem by comparing and using knowledge in creative way to find a solution to a problem. And again today the knowledge is not complete. Human brain can hold a million times more knowledge, and compare it in a unique creative way. But this robot can’t. This robot has only basic knowledge, and can help you by reminding it to you, and find a solution FASTER.
Do you want to know how to check if the robot doesn’t have full information? Just ask something you have massive knowledge. Do you want to know if the robot can generate something or find a solution? Ask to make up a joke.
I asked: “Make me a joke about skirt hemline and a needle” ChatGPT gave me this: “Sure, here’s a joke for you: Why did the seamstress have a hard time hemming her skirt? Because every time she tried to use a needle, the hemline kept running away from her!” And another one, which is worse: “Why did the fashion designer need a needle to adjust the skirt hemline? Because he didn’t want to be left hanging by a thread!“
And that is a perfect proof that this robot can’t propose to you a unique and perfect compilation of different texts.
I experimented with writing an article with ChatGPT. I ended up writing the article like always myself, although some details I forgot about the subject and reminded me by this robot was useful as I didn’t need to take a long time to search for them.
This AI image generator MidJourney got a massive fright. Everybody saw the Nike / Rococo sneakers images on Instagram as if they were real, or weird Balenciaga / Harry Potter stunts on youtube. But did you check the manual? Did you check how this robot actually works? Well, I did.
Did you ask yourself why this MidJourney “lives” in the chat? Well…. pose for dramatic effect… Because this robot also WORKS ON THE TEXTS, and not on the images.
In fact, when you upload an image to this robot, it describes it in the words to itself. This robot has basic knowledge about blue eyes, and leather skirts for example. It describes it in every detail, and then draws you something based on the text description it has. This robot has a huge base of descriptions of the styles, faces etc…
So all these copyright infringement issues… go to the window, as this robot only describes the details of the painting or photo, makes it like a text, and then generates something based on the description and not on the image.
Here are the examples for you. Try this robot – it’s actually fun. Try to upload your own photo and add some description to it, ask to change the outfit or background. First of all, the face – that won’t be you. It would be someone else’s face, with probably the same features like blue eyes, or button nose. But it won’t be you.
Then look closely at the fingers. This robot still doesn’t know how many fingers humans have. You can get 6-7 fingers… even funnier, you can get 3 legs or 4 hands. I got them when I experimented with my own photos.
Yes… On both photos that was suppose to be me. This generated portrait by MidJourney doesn’t look at all like me. It has 4 hands, and some hands have 7 fingers. Pretty scary… And on anther photo with huge rabbit it’s also not me, and this woman has a serious problem with her legs…
So when you see my avatar we recently presented – that was Midjourney, and lots and lots of Adobe Photoshop filters and retouching afterwards.
We recently presented two imaginative projects. The concepts Chanel with Lenotre, and Louis Vuitton with KFC. Do you think that these images were entirely generated by this robot? Absolutely NOT. I had an idea, I developed the concept and the details of it, I used MidJourney. And then we had 7h of Photoshop afterwards.
Then word or logo. Try to use your own logo in MidJourney. The results will be smashing – not at all your logo, the text is even worse. This robot can’t replicate exactly your image logo, and can’t put exact word on the picture. Why? Simply because the description it does of your logo is not precise enough!
I tried with the logo and the word RUNWAY. So instead of RUNWAY I got this… no matter how many times I tried. There’s no way to get the word.
And here we are talking about fast creation? or copyright infringement?
I asked: “Louis Vuitton logo on the futuristic building, on the background futuristic garden“. And that’s what I have…. no logo, no Louis Vuitton name spelled correctly. I don’t see any “fast design” opportunities.
So the mystery is solved. Midjourney can more or less understand the colors and the structures, has basic knowledge about things like logos, or magazine covers, it can describe them but it can’t replicate them, it can’t create.
And I’d also add a detail here. Try generating an image with MidJourney and then enlarge it. You don’t see the pixels like we usually do when we closely look at the photos, you see the lines. This robot draws by the lines. It is another proof that this robot doesn’t take images, and regenerates them into something else. It draws from scratch.
So my point is, which should remove all fears and panics related to “fast creation” or “fast design”, using MidJourney AI image generator can only help to create. Human create the concept, human create the idea, without it MidJourney or any other AI robot is useless, it would never create a great art piece or concept.
So I invite you to try these innovations – AI technologies, they can be handy sometimes. And I also would like to remind everybody a story about the first Photoshop when it came out on the market. This new software at that time made a huge panic… but since then we use it to create great photos and art pieces. It is a tool, as AI image and text generators are.
So let’s try it together….