Google has launched its AI Experiments website today.
AI Experiments is the brainchild of mostly millennial coders who work in its Creative Lab in New York. AI Experiments showcases simple experiments intended to convince people that AI isn’t just for the academia.
Outside of the tech community, people view machine learning and artificial intelligence as a subject that is interesting only to people with PhDs. The AI Experiments website offers five demonstrations of AI, so people can fool around with examples of machine learning. One can even download the code to see how it works.
Google AI Experiments for Fun
As of now, there are eight demonstrations available to interact with:
This experiment lets you make music with the computer using a picture. The algorithm uses image recognition to label what it seems, and then the labels into lyrics of a song. Girgio Cam was built using MaryTTS, Tone.js, and Google Cloud Vision API.
This game is built on machine learning. With Quick, Draw! you draw, and the neutral networks will guess what the object on the screen it. The more you play on it, the more it will learn.
The Infinite Drum Machine
This experiment is quite complex. Basically, it uses machine learning to organize everyday sounds. With the help of the technique called t-SNE, the computer put similar sounds together. One can also map to explore neighborhoods of similar sounds.
This experiment is perhaps the most beautiful example of what you can create using Google’s machine learning API’s. Basically, it lets you take a picture of something you hear to say it in a different language.
Bird Sounds uses machine learning to organize bird sounds. Similar to The Infinite Drum Machine, it uses t-SNE, to place together similar bird sounds.
This neural network is trained on musical concepts, melodies, notes and timings. It lets you make music using machine learning, wherein you just need to play a new notes to see how it would respond.
Visualizing High-Dimension Space
This experiments lets coders visualize what is happening IRL in machine is learning. It allows you to see and explore high-dimensional data.
What Neural Networks See
Another interesting example on AI Experiments is What Neural Networks See. It allows you to turn on your camera to explore what neural networks see using your camera!