That system could stay as-is if the normal crew would be automated away.
There's also the question of who's going to deploy (and untangle) the boarding ladder that pilots use to get aboard vessels. As well as the prospects of hacking that subsystem in order to divert the vessel.
Now design me an automated deployment system which can handle the operating conditions and environment, and respond, effectively, as good or better than a human can.
Consider that your ladder can get snagged, caught, swing in the wind or seas, collect flotsam or jetsam, etc. Things can get very squirrely at sea.
Sometimes humans aren't the best at a specific function, but through processing and manipulation capabilities, including the ability to use and apply multiple other tools, they're the best means for responding to a general case of circumstances which cover a lot of ground.
Not hard to imagine something better than jumping to a ladder on the side of the boat to be honest.
Which is already the case, so this would be a fairly easy transition.
You really think computers of this day and age are less capable? We trust them for a lot more risky stuff than piloting a ship at very low speeds.
It's only when things get crowded and unpredictable -- that is, with lots of non-automated shipping about -- where you will need serious piloting software. I imagine that there will be a long crossover period where ships will set out from a port with a small human crew -- possibly just a pilot and lookouts -- then go autonomously across the ocean to be picked up by another small crew a few hours out from the destination port.