I'm not debating that it can be done, and it works. I just feel like it contributes to the "dumbing down" of drivers, as I feel automatic headlights/wipers/tpms do.
Then again, maybe the whole concept of "dumbing down" drivers is oxymoronic, considering some of the things that happen out on the roads.
I guess I'm just a bit of a throwback - I prefer the old "mark 1 eyeball" on a dipstick for oil/trans fluid levels, I know how to use a tire gauge, and I like having a mechanical level for selecting gears (and actuating the transfer case on my Jeep).
(edit) Maybe my problem with going "dipstick-less" is that I have no way to monitor the levels myself, and the sensor is just a boolean - good or not-good). I have no way to see for myself what's going on, or even verifying that the sensor is reading correctly (or at all!) - I just have to trust the computer. Being a Software Engineer by trade, and specializing in embedded/embedded test systems, I know that just because you write the code to do something doesn't mean that it actually will the first time you run the system - there's all sorts of ways they can go wrong.
That probably is part of my aversion to self-driving cars - given some of the things I've seen SW people do, the concept of software being responsible for driving a car is, frankly, occasionally terrifying to me. Now, if they decided that all self-driving car SW had to meet DO-178/278B and DO254 (FAA safety-of-life-critical requirements for software and firmware/FPGAs, respectively), I might be at least slightly less uncomfortable about it.