People have a new hobby. On their social media profiles, they no longer just share the usual moods and what they had for lunch on vacation, but suddenly seem to have developed an almost overflowing imagination. Suddenly you see images of ants floating in space, cute marsupials in suits or futuristic machine cities. Where does the creative explosion come from?
AI is not on the rise, it has been here for a long time. Just high-end, now mass-produced. The generators for images (Dall-E2, Midjourney) and text (GPT-3) from scientific papers have made it to the end user within a very short time. The promise: In a few keywords you describe what you want to see on the desired image, the machine does the rest. With some you still have to manipulate the controls and settings yourself. At others like Midjourney is enough via the Discord chat platform Giving orders what to see. The bot then serves four suggestions within a minute. In test versions it is free, but the image creation costs 15 dollars from a provider, for which the AI generates up to 460 images.
So there are no longer any limits to the imagination. Anything you can think of, the AI can create. The question remains as to what the user’s imagination is actually like.
Not very good at first glance. Or why else has some kind of secondary industry developed around the image and text creation platforms? In marketplaces like prompt base, Great prompts or neutron field the finished illustrations or texts are not put up for sale, instead the users only trade with the keywords that the AI has to give in order to achieve a good result.
Are people really too lazy to think about what they want?
The so-called prompts are ultimately work instructions for the AI. What style should the picture have? Which author serves as a role model? Which color palette can be used? Pastel or rather neon? Portrait or panorama? And last but not least: What should actually be seen?
The prompts are not expensive. Most users ask for single-digit dollar amounts. But why should you buy them at all? Are people really too lazy to think about what they want? Luckily it hasn’t gotten that far yet, but because you’re dealing with an AI on the other side, it’s not enough to write down the prompts in a reasonably reasonable syntax. A successful work instruction, on the other hand, is more like this: futuristic city under a dome digital art deviantart high detail high definition octane render. You have to know exactly what you want – and how to formulate it in such a way that the machine gets it.
While the Twitter layperson may end up with a blurry nightmare, experts have long been using the programs to produce professional-quality illustrations. Album covers, company logos, even entire comic books have already been developed by Dall-E2. The question of copyright is interesting. Who actually owns the rights to the images created in this way? It can’t be the software, because otherwise all images that have ever been edited with Photoshop would have to belong to the Adobe group under copyright law. So can something as supposedly incidental as the exact formulation of the work assignment be the intellectual achievement worthy of protection?
A lot of people around the world are scratching their heads about this. Meanwhile, it almost seems as if you can witness the emergence of a new job profile: the AI whisperer. In other words, a person who manages to elicit the desired effect from a computer by using apparently incoherent words. Mankind has never been closer to actually working magic spells.