In 1953, J.C.R. Licklider, a computer scientist at Harvard, created the world’s first speech-understanding machine.