Human infants are the best learners in the universe - curious, self-motivated, and able to learn unsupervised from much smaller, uncurated data. Efforts to simulate their knowledge acquisition in machine learning have already begun in academic research.

 

I was curious to explore the potential of infant cognition in AI's development, but wanted to do it visually. I therefore made a data visualisation of a bot trying to learn like a baby. In the website, a web scraper gathers information on artificial intelligence from Google Scholar. Using a learning mechanism in infants, it tries to map similar information and understand this data. As the bot processes more information, it learns about machines and henceforth its own identity along the way.​