i tend to say that AI is an immature thing, i.e. it can turn into a great thing if it’s properly developed, which today by far it is not.
things that would have to change, for example, would be using only training data where the license actually allows it. another example is reducing the extreme energy usage of AI systems. another example is better theoretical background understanding, like, under what conditions does AI output stable, reliable output? what mistakes does AI make, under what circumstances do they appear and how dramatic are they?
i tend to say that AI is an immature thing, i.e. it can turn into a great thing if it’s properly developed, which today by far it is not.
things that would have to change, for example, would be using only training data where the license actually allows it. another example is reducing the extreme energy usage of AI systems. another example is better theoretical background understanding, like, under what conditions does AI output stable, reliable output? what mistakes does AI make, under what circumstances do they appear and how dramatic are they?