Software developers are keen to show off the latest in artificial intelligence, which is why this technology is coming out of the labs in the form of online tools that anyone can access through their browser.
Whether you want to generate weird and wonderful AI images from text prompts or create a musical composition in partnership with a computer, there’s plenty to explore.
These apps are getting better and better over time, and they can give you a good idea of what AI can do and where it might be heading in the future.
This person does not exist
This Person Doesn’t Exist produces AI-generated images of faces that look incredibly real, but they’re not. Just load (or reload) the page to get a new one.
It’s the work of software engineer Philip Wang, and it shows the power of publicly available Generative Adversarial Networks (GANs). They are dual systems where one set of algorithms or neural networks produces an output (in this case, an image), another evaluates its quality, and they repeat the process millions of times.
[Related: Artificial intelligence is everywhere now. This report shows how we got here.]
In this platform, the first neural network generates a face from a large library of data. Then, the second neural network evaluates the realism of the face based on similarities to images of real people, and it only accepts the image if it exceeds a certain threshold.
While this tool can ensure that the world will never run out of generic photos again, it has other, more sinister implications. Among these is the possibility of creating identity documents for people who do not actually exist.
If there’s an artist in you, then Magic Sketchpad could help bring it out. This is an AI experiment from a team at Google that gets a neural network to walk you through. Every time you drop a line, the platform will respond to your scribbling by finishing the drawing according to a set category.
The neural network was trained on millions of doodles taken from the highly entertaining Quick, Draw!. Start by choosing a category from the drop-down list at the top right of your screen – there are plenty available, from frogs to sandwiches. Magic Sketchpad knows the kinds of shapes and lines people tend to make when trying to draw simple concepts like a bird, boat, or cat, so it can predict what you’ll draw next and finish. the doodle for you.
The tool can also help artists enrich their work or provide new incentives for creativity. Maybe one day we could see computers doodling as well as humans.
If you’re more of a musician than a designer, then AI Duet might be a better fit for you. Built by a Google engineer, AI Duet places a keyboard at the bottom of your screen and produces an auto-generated response based on what you play on it. You can click keys on your screen, hit them on your keyboard, or even connect a MIDI keyboard to your computer.
A traditional approach to a project like this would have involved a programmer coding hundreds or even thousands of responses to specific patterns that a user could play. But AI Duet comes up with its own answers based on a huge database of songs it’s trained on. This gives the program the ability to generate melodies that match a user’s input without any specific instructions.
This is another example of how AI can work in tandem with artists to produce new creations, whether for movie soundtracks or background music in games. Theoretically, you could rework a riff an infinite number of times.
If you’ve been on Twitter lately, chances are you’ve seen the creations of the Craiyon AI Image Generator, formerly known as Dall-E Mini. It’s basically a neural network that turns text inputs into images: you type in what you want to see and the system generates it.
To generate images, Craiyon extracts information from millions of online photos and their captions. This means he has a vast visual knowledge of everything from celebrities to national landmarks.
The results produced by Craiyon are a bit rough at the moment, but it’s not hard to see how we could possibly use this technology to generate very realistic images from scratch using just a text prompt.
Even stranger things
Even Stranger Things is worth checking out even if you’re not a fan of the Netflix show that inspired it. The platform allows you to submit a photo of anything you like and turn it into a stranger things-poster style.
The site was built by creative technologist David Arcus, and it leverages the Google Cloud Vision API, a machine learning system trained to recognize images based on an extensive database. Thus, by processing thousands of photos of dogs, for example, the AI learns to more accurately spot a dog in other photos.
Even Stranger Things will try to identify what’s in the image you submitted and incorporate it into the final design, usually with overall accurate results.
It’s a fairly simple AI tool, but it shows how we can use databases to teach machines to spot new patterns that aren’t in their training materials. The platform is also a great example of how algorithms can apply a particular visual style to photos to create something new.
talk to books
Talk To Books is another artificial intelligence tool created by Google engineers. In this case, the platform uses words from over 100,000 books to automatically answer a question or text prompt.
Although you can’t really hold a conversation with the site, you can ask questions like “how do I fall asleep?” and “how did you meet your partner?” to get answers that make sense. Type your prompt, then press Go to see the results, which you can filter by literary genre if necessary.
This is another example of how machine learning allows AI to predict a correct answer to a question or prompt by analyzing patterns in text.
At this point, the AI can’t really finish novels or even news articles, but with enough data and refinement, those are certainly possible uses for it in the future. We might even be able to complete notoriously unfinished works of literature just as the original authors would have done.
As the name suggests, Pix2Pix is an AI image generator that takes a photo and transforms it into another. In this case, the tool shows you a photo based on something you doodled.
Scroll down the page and you’ll see there are four different examples to try: cats, buildings, shoes, and purses. Sketch your design in the left window and click Treat to see what the AI does with it.
This is another GAN-based engine, where two neural networks work in tandem to produce realistic results, and even determine where the edges of objects in images should be.
Turning sketches into realistic photos can be extremely useful in all sorts of fields, from constructing buildings to designing video games. And the quality of the results will only get better over time as these neural networks get smarter and smarter.