Considering the limitations causing a person harm would put on a mission, and the complexity of bringing them back to Earth again, I'd have thought it'd be more flexible to send a bunch of robots.
Launches are rare, which means you want to pack as much value generators into your payload as possible, which now means your payload is an expensive one-off workpiece (vs. a mass-produced item), which means extra risk, which means more safety requirements around integration and launch, which means more expensive launches, which means launching less often, ...
So if you're thinking of sending something to Mars for Science, people look attractive (mass produced, can do a lot per dollar spent on getting them qualified for mission), but life support requirements make this too expensive, so sophisticated one-off robots fly from time to time instead. Now if we could increase launch cadence and payload capacity, which is what SpaceX is working on, we could start sending people... or we could load the Starship with multiple cubesat launchers and a payload bay full of remote-operated or semi-automated rovers and drones. In few missions, we could blanket most areas of immediate scientific interest with robots. In this reality, higher payload capacity and launch cadence would drive demand, which means both the scientific robots could be mass-produced, and there would be more iterating on design, increasing the amount of missions that could be done by machines.
The way I see it, right now what SpaceX is doing to enable manned missions is pretty much aligned with what's needed to enable proper robotic missions.
Except if you mean T-1000 or something...
And because they eventually want go to Mars, have a little scientific colony, etc?