I begin to ponder whether it would be much more effective to ask MN these questions, someone will know the answer. Maybe that's why the AI companies are trying to scrape it.
@Lalgarh In theory, I do not have anything against people winging it, but there has to be the right situation and context for it. Something that doesn't affect anyone else if you get it wrong and which is not that important in the grand scheme of things - you can use less effort and try to make it. However, something relevant and which has to adhere to laws and could impact others, like tax offices, hospitals etc - no improvisation on the fly.
I would say AI is more like interns who just don't care and take everything literally - yes, they did the task, but with the lowest level of understanding and care about the results.
When it is pattern recognition for useful applications, which do exist, too, I think this should not be called AI, but just an improved program compared to before. Then it is clear these are machine-learning results and they always need oversight.