The two fronts of computer learning

Have you noticed that OpenAI is focused on intelligence that creates new content, be it pictures or explanations (gpt3)

While most of machine training efforts are focused on changing existing truths?

If a person is walking and you tell the computer to make that person be a different thing (a dog, a mailman, or a president) you are changing the truth

But if you say to the computer "Do a scene where a person walks" you are creating the truth