I think we have a set of important lessons for the future of AI over the next decade. The first is its source, a team of three looking for an idea in a random corner of the OpenAI building.
This single idea about the diffusion model, just a little breakthrough in the algorithm, let us go from doing something not so good to something that can have a huge impact on the world.
Another interesting thing is that this is the first AI that everyone is using, for several reasons. But one of them is that it creates a complete finished product. If you’re using Copilot, our code-generating AI, it must have a lot of help from you. But with the DALL-E 2, you tell it what you want, like talking to a fellow graphic artist. I think this is the first time we’re seeing this in artificial intelligence.
3/ What does DALL-E mean to society
When we realized that the DALL-E 2 was going to be a big thing, we wanted it to be an example of how we were going to deploy the new technology—to make it clear to the world that the images could be fake, and it was like, “Hey, you You know, pretty soon you’ll have to stop trusting pictures on the internet.”
We also want to talk to the people who will be most negatively impacted first and get them to start using it. It’s not the current framework, but as a field, I hope we can get to a world where if you help train an AI by feeding it data, you should somehow own part of that model.