Typically, AI systems require huge amounts of data to mimic human speech. A new study shows that it may be easier.
When Sam was six months old, scientists put a helmet camera on him. From then on she recorded the Australian toddler playing with his brightly colored toys. How Sam notices the cat that suddenly appears next to him. How he sits on his mother’s lap and looks through a children’s book with her. As she explains something to him, Sam slaps his hands at what he sees in the book. During the recording, a microphone recorded what the people around Sam were saying. Over time, Sam began to pick up words and give them meaning.