Not my views personally, but there was a time on Twitter when doubting the effectiveness of certain drugs, expressing views about the 2020 election could get your account suspended. I think people are hoping that things like this won't happen any more because of Musk's views on free speech.
Maybe, the insurer would still needs a claims-handling staff overhead to verify a claim made is valid. And perhaps also an underwriting team to assess the risk before the start of a policy. The DAO could replace the paperwork/emails, but I imagine that is only a small expense.
The ML skills update was a very basic update. Proper ML skills take a long time to learn and can't be taught through an exam, the material for which gets forgotten very quickly afterwards.
The incumbent insurance companies maintain their position in the market through having large amounts of capital which lets them ultimately offer lower premiums than a new competitor entering the market which may have less capital. This means the legacy insurance companies don't need to prioritise investing in innovation to maintain their position, hence the old companies stay around.
Actuary here, having previously studied maths. I considered the exams hard not because of the level of understanding needed, but because of the amount of content that needed to be learned for each exam. I think most people studying A level maths can pass the exams.
Qualified actuary here also from UK. You're right about insurance being a legacy industry. In terms of the exams, if you can pass A-level maths, you can pass the actuarial exams. I found the main reason they were considered hard is not because of the subject matter, but due to the amount of content that had to be learned in a short space of time.