His universe Artificial Intelligence reaches higher and higher and unexplored peaks. Today, many IT products, such as cameras of our municipalities smartphonessearch engines, the software we use every day, smart products for home automationand so on.
Thanks to his concepts Machine Learning Electronic machines can learn a lot from us by imitating our behaviors and are also pushed into the creative realm. This is the case with some amazing new programs, capable create images almost alone.
Obviously, it could not lack appeal in this area as well Google, the IT giant that strives to always be at the forefront. It develops a software called Imagen, which aspires to turn text inputs into very high quality images. Even in his case creative and wacky requests.
All this thanks to Artificial Intelligence and the utilization of huge database material on the internet. Will he be able to be efficient but also ethical and not indifferent? Will it do better than competitors who expected it to? Let’s see what it consists of and try to find out.
Google creates Imagen: what it is and how it manages to create
Imagen is a fairly recent project from Google, which has the (definitely not simple) goal produce very high quality images starting with a simple text description.
Simply put, he wants to give us the opportunity to write something like this “A dog in a cowboy hat surfing a lake on a rainbow board” and get his picture… with a dog, in a cowboy hat, paddling across a lake, on a rainbow surfboard.
All this in one high visual quality, as if we had created a special illustration in hours and hours of work between drawing and coloring. All this as if we had hired a specialist photographer and he had at his disposal a complex photo set, consisting of many environments, many people and animals, many backgrounds. All these really and without cost.
Such a commitment is possible by going through the aforementioned machine learning in artificial intelligence systems. The software should proceed with the written instructions one step at a time, effectively combining them because it draws from a multitude of images that have been “curated” to it from the web.
In practice, it is a matter of giving as much data, information and images as possible “In meal” at Imagen. This is how it works in this field: the software is fed with a huge amount of data so that it can learn from it and merge it through the concepts they contain. That way, when we write our imaginative description, the program will match everything, it will go through thousands of images and it will give us a quality result.
The intention of the big G is to become a giant in the creation of images as well, which it already does in several unexpected areas, such as the Drive cloud storage services or the very recent Google Pixel smartphones.
When and how you can use Imagen: between problems and competitors
Google has made one available presentation website for Imagen full of brilliant, unexpected and incredibly detailed images. But if you’re wondering where to try the software to get a nice picture of the dog on the surfboard, the answer is unfortunately. negative.
The software is still only in development, and is not yet accessible for public use. There is also no release date: we will have to wait, no one knows until, to create images of any kind.
In particular, we focus on the possible moral issues of such powerful software. It is true that the first thoughts go to the photos buffet or what could be useful for day-to-day operations, without having to design (or create) it ourselves.
Imagen, however, could receive commands that are not at all nice and nice, sometimes even completely unpleasant the illegaland in this case Google He does not want that the software gives its usual satisfactions. Furthermore, even with innocent or well-intentioned elements, the program is important you are not wrong and does not return creepy, disturbing or disturbing images.
In fact, it is a delicate issue that is fully involved morality. As long as we talk about funny photos, landscapes and objects, we all agree. But when some images may offend some groups of peopleGoogle knows you need to be careful, so development may take some time.
There are a few on the market right now antagonist who expected Imagen, like the well-known Dall-E Mini, used to create cartoons and memes on the network. Of course, however, the images generated by this software are of very poor detail, and have so far only been used for entertainment purposes.
It will take time for AI to reach greater heights and as usual, Google wants to be the strongest. But we’ll see if it succeeds: for now we can just sit back and watch (or think we want an image of an ancient art museum submerged in water navigated by little robots, as a preview from the company showed us).