Experimenting With GPT-3 Felt Like Witnessing a Technological Revolution
The potential for the new artificial intelligence program is nearly unfathomable
On July 3, 1896, the Lumiére brothers embarked on one of the most remarkable demo series of all time. After creating one of the world’s first films — a 50-second clip of a train arriving at La Ciotat Station in Southern France — they toured the world, projecting their new creation for audiences who had never seen a film before. As the train approached the camera, many people reportedly (and perhaps apocryphally) leaped out of their chairs and dove to the floor. They thought the train would come out of the screen and crush them.
After seeing the demo in Russia, critic Maxim Gorky realized, correctly, that film would forever change the world, writing, “Last night I was in the Kingdom of Shadows. … The extraordinary impression it creates is so unique and complex that I doubt my ability to describe it with all its nuances.” (The “shadows” were Gorky’s attempt to describe the appearance of flickering black-and-white images on a screen.) Grappling for words, Gorky wrote that the new medium of moving pictures was “strange,” “terrifying,” and ultimately “edifying.”
This week, I received access to GPT-3, a landmark artificial intelligence technology from OpenAI. GPT-3’s utilitarian name belies its incredible potential — for good and bad. I’ve seen a lot of technologies, and I’ve done a lot with A.I. after working in the field for over a decade. I can say with no irony or hyperbole that GPT-3 is the most important technical object I’ve seen since the internet itself and certainly the most significant artificial intelligence technology created in this millennium. Testing it out, I feel a bit like Gorky descending into the “Kingdom of Shadows” — at a loss for words as I bear witness to something that is simultaneously totally new, deeply terrifying, and completely electrifying.