For example, in 2019 he said “Human radiologists are already much worse than computer radiologists. If I had to pick a human or an AI to read my scan, I’d pick the AI.” [0] However, most people with domain knowledge in this space still would not want their scan not to be read by a human. The AI radiology apps are generally narrowly focused and useful for consultation not diagnosis. [1]
In 2015 he predicted: “Self-driving cars are going to get here much faster than people think,” Altman said. He thought we’d see them in three to four years. [2]
I think he would be wise to learn from Bill Gates that it's a bad idea to over-predict what will happen in the short term. "We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten."
I'll add to this that it's generally a bad idea to make big predictions outside of areas that you have strong domain knowledge in. I doubt Sam Altman has strong domain knowledge around US higher education. Unfortunately, Silicon Valley culture is full of people who feel that because they've succeeded in one domain they have some kind of hidden insight into others.
[0] https://www.cnbc.com/2019/02/26/sam-altman-on-ai-jobs-may-go...
[1] https://link.springer.com/article/10.1007/s00330-020-07230-9
[2] https://techcrunch.com/2015/10/06/elon-musk-sam-altman-say-s...