AI has focussed on algorithms for robots to mimic human behaviour. Imitation is prevalent for young children, but for teenagers and adults much learning is by trial and error, insight and critical thinking. They solve problems, make comparisons, judge situations, synthesise and create. AI would have to recognise what situation to imitate with its algorithm, without dumbing itself out of business. If you want smarts, use a human.
When a claim is falsifiable, Karl Popper had the burden lying with others to try and falsify it. If it is not falsifiable, Bertrand Russell had proof lying with the person who made the unfalsifiable claim. Could this mean a researcher cannot gather data to verify their thesis unless it is unfalsifiable? My paper explores proof in climate change. Greenhouse Teapot 040717