Part of the reason actuaries make so much is that they are certified and legally required to sign off on analysis, similar to the way civil engineers put their

Part of the reason actuaries make so much is that they are certified and legally required to sign off on analysis, similar to the way civil engineers put their signature (and reputation) on their work. But you would be surprised how many decisions come down to "actuarial judgement", having good business sense, intimate knowledge of insurance rules and regulations...and not doing anything special with the data. Hell, some actuarial methods are the same as they were in the 70s. Data science is slowly eating their lunch in predictive modeling areas, but I wouldn't be surprised if they're still around after data science itself has been automated. I work on a team of actuaries in a more data science-y role. Believe me, I had the same impression initially: "Can't we automate some of this stuff?". But I've since come to see their value.