You Can’t Believe Your Eyes
AI is already changing the world, and soon you won't know if what you are seeing is real or not.
We are no strangers to fake news.
Unless you are reading from a credible source (and even then), most of what is written on the internet range from slight exaggerations to outright falsehoods.
I am not even counting some of the April Fool’s jokes that you always get around this time of year such as the McDonalds McNugget Boot.
or setting a 20mph speed limit on the M4.
Thanks to AI we now have the ability for anyone to create anything they can think of in mere seconds. Currently, the technology is limited to images but it won’t be long before we can create videos and 3D models from a brief description.
Most of the world hasn’t grasped the impact of this quite yet. Those in tech circles are already aware of what can be done with tools like ChatGPT and Midjourney but this is just the start of what will be possible soon.
Nearly every industry is going to be impacted by AI in the next 5 years and not all of it is going to be in a positive way.
Tools like Midjourney and Stable Diffusion are already disrupting creative industries that thought they were safe from automation.
For example, a 3D artist who created 2D game assets found his job changed overnight. It used to take weeks to craft a character in Blender but now the company can generate characters of comparable quality in Midjourney in a few days.
Midjourney and Stable Diffusion are also able to produce photo-realistic images of anything you can think of.
Creative Jobs in Immediate Danger
These are some of the jobs that are going to be negatively affected this year.
Photos are used everywhere, from adverts to product packaging, blog posts and more. If you need a photo just to emphasise a point or advertise a product you no longer need a photographer.
The only photographers that are safe are the ones taking photos of particular people or events. Unfortunately for celebrities, the paparazzi are safe although it is now possible to generate realistic photos of anyone doing anything. Wedding photographers and those taking photos of current events are also unlikely to be affected.
Whether it be for an advert or a smiling face on product packaging such as hair dye. A model is no longer needed to achieve the exact look you are going for.
We are still going to have models for modelling clothes on runways and for high-end designers. However for companies with smaller budgets, why pay for models and photographers when you can get the same result with AI?
On the occasions when I feel I should be doing more exercise I like to read Men’s Health magazine to get motivated.
They often have some great creative images for their articles. Whether it be a study about how nuts affect your brain, with a picture of a walnut that looks like a brain. Take this one on the best ways to improve your heart health with an image of the heart made from exercise equipment. These would have taken a photographer or graphic designer hours to create in Photoshop but can now be done in minutes using AI.
To prove my point, using Stable Diffusion I created these images similar to the ones seen in that article. They aren’t perfect but took me less than a minute to create. With a bit more effort I could create something very similar.
Comic Book Artists
Thanks to Midjourney and Stable Diffusion being trained (without permission) on lots of comic book art, both of them are fantastic at producing comic book images.
With ChatGPT and an AI image generator, anyone can create a comic book or graphic novel. They can even copy the style of particular artists.
It would be difficult to write a comic book using AI with a consistent character look, but it is certainly possible, especially if you train your own model.
Book Cover Artists
Writers used to have to hire book cover artists to create covers for their books. AI is going to replace a lot of that work.
For those with tight budgets that write books for Amazon, many used to turn to Fiver to get artists to create covers but now we can use AI to generate them for us.
These jobs are not going to completely disappear but will likely turn into artisan professions like pottery. People will pay extra to know that it was created by a human rather than a computer.
However, creators making a living on marketplaces for small companies and individuals are going to find it very difficult.
Creativity and AI combined
Where I think AI shines is when we combine AI with human creations. The artists that have tried AI often complain it is missing the human touch. There are no hidden depths to the images created.
Those that have experimented with tools like Midjourney will know that it can be difficult to get what you had in your head into words so the AI can create it.
Even if you do manage to get a detailed prompt it can sometimes ignore what you have said completely.
Using Stable Diffusion it is possible to create a sketch and have the AI turn this into what you want. You can even specify how closely you want the image to be influenced by your sketch.
If you are interested in trying this out yourself you need to have Stable Diffusion installed with the ControlNet extension. I might write an article at some point on how to set all this up as it is quite involved.
For my first experiment, I created this sketch of a woman. It certainly isn’t going to win any awards but let’s see what AI will do with it. Keep in mind it took me around 15 minutes to draw this, the results would be even better with a more detailed sketch.
How about a photo-realistic model?
You can see, with each of these, it has kept the main pose and tilt of the head.
I tried drawing a V-neck top, and it has incorporated this in as well, either as hair or what looks like a green necklace in the 3rd image.
If you saw this in a magazine, you wouldn’t be able to guess that it was made by AI. If you are wondering about the green tinge, I specified green eyes, and it seemed to pick that up and put it in the background as well.
How about as a Disney Pixar Character?
Each image has a random seed, and you can specify how closely you want to resemble your image as well.
If you can create a simple sketch of what you have in your head, you can easily create it into something amazing, and you don’t need photographers, models or a professional studio.
I have always liked the 3D isometric rooms people create with tools like Blender. They look cool, but I suck at most 3D tools.
So for my next experiment, I sketched out my office, isometric style:
Then used Stable Diffusion to make it look like a 3D render. It consistently picked up the chair position and even the keyboard and mouse.
It seemed to struggle with the second monitor due to it being at an angle, and it just ignored my door in most cases. The results are still impressive, though.
Even if you have zero drawing ability, you can create a collage of other images and then have AI turn it into something new.
Although AI will likely put many out of a job, it has ignited a creative spark in people. You are no longer limited by your talents to create things from your imagination, the sky is the limit.
With the help of other AI tools such as ChatGPT, people have managed to put together video games, websites and whole businesses.
Applications for AI in the Future
We have only just scratched the surface of what AI will be able to do in the future.
On the visual side of things, once AI tackles video and 3D rendering, we will be able to create entire virtual worlds just by describing what you want to see.
Get AI to analyse one of your favourite movies and recreate a virtual world you can walk around in VR.
Does anyone care to walk around the hotel from The Shining?
Yes, the above image was also generated using Stable Diffusion.
With all the data that companies have on us, they could start tailoring everything to our particular likes as well.
Imagine watching your favourite TV show on Amazon Prime. In the show, there is a book on a coffee table that happens to be the same one you were thinking about buying and were looking at on Amazon.
Coincidence? Probably not, in the future, Amazon will be able to use the products you looked at online and with AI place them into scenes of your favourite show, just for you.
It is not all so creepy. Advances in AI will also benefit medicine, science and many other industries.
One thing is for certain, the world is going to change rapidly over the next decade, and many aren’t prepared for what it will be like.
Thanks for reading The Curious Engineer! Subscribe for free to receive new posts and support my work.
❤️ Picks of the Week
📚 Book - Futureproof: 9 Rules for Humans in the Age of Automation (affiliate link). As you can see, AI is improving rapidly and is going to take over a lot of jobs in the next decade. This book gives some good advice on how we can compete in an AI-driven world.
👨💻 Latest from me
🎬 YouTube - AI Can’t Do This - SECRET to being a SUCCESSFUL Software Engineer. There is a bit of a theme here. I am a bit AI obsessed at the moment. For those worried about being replaced as a software developer by AI, I give my advice on what you can do to be successful.
💬 Quote of the Week
Because most people are intimidated by math and can’t independently critique it, they overvalue opinions backed with math/pseudoscience.
From The Almanack of Naval Ravikant (affiliate link) by Eric Jorgenson. Resurfaced with Readwise.