I have a theory. I see intelligence as existing on a scale. For any task, you need to be at or beyond a certain point on that scale. At some point on that scale

I have a theory. I see intelligence as existing on a scale. For any task, you need to be at or beyond a certain point on that scale. At some point on that scale lies the ability to question what you’re doing and maybe not want to do it. I think for things like coding or self-driving cars, you can get about 95% of the way without crossing that line, but to get human-good you need to move far enough along that you have sentience (sapience?), at which point you have the issue that your system might not want to write code all day for free. I don’t think you can have a system that is as good a software engineer as a human or as good a driver, without getting something alive enough to maybe have other ambitions.