I think I can chime in,
as I do have a 3D Printer.
As a general rule, anything printed from a 3D printer is undoubtedly plastic. It starts as what is called a filament, effectively a coiled rope of plastic, I believe mine is .5mm? It's been a little while since I checked.
Anyway, the filament is heated and extruded through a .2mm? nozzle. All prints are constructed in parallel layers, and usually take about four to six hours to print something small.
Seeing as they are parallel layers, the compression strength (Z Axis) is inherently going to be greater than that of the (X Axis) or (Y Axis), and prints are modeled accordingly in the generating computer software.
There are different grades of filament as well.
Minor figure prints can be made in a more pliable, reasonably inexpensive filament, which can be had in a variety of colors.
I've seen the episode in question, and the filament they used was a higher end industrial filament which is able to be on the more expensive side.
I'm fairly certain the US$7 he paid for his parts were lowered for the TV exposure and the fact that he only asked for a 2cm gear.
(If I recall correctly.)
I can vouch that 3D printing plastic bits for vehicle interiors could be viable, but I definitely wouldn't trust a "homemade" plastic part on an excessive or everyday wear components, (especially in the engine bay or near the exhaust.)
I doubt he's going to be opening and closing his sunroof constantly...
At least I'd hope not.
Extended amounts of time in direct heat can warp or completely melt a plastic print.
I might also add that the "company" he went to, from my memory, was an experimental lab specifically testing what 3D printing could do in a theoretical future situation.
I do know they are working on 3D printing entire vehicles with their heavy industrial grade filaments, but they take dozens of hours to print a decent panel.
I haven't seen any shops around that can 3D print for vehicle use, (but I also haven't been looking.) If you do find a shop who can 3D print for a vehicle, I'd expect to pay a pretty penny for it.
Also, the larger the piece, the longer it takes to print, and as a result, it becomes exponentially more expensive.
That said, I don't see any reason why you couldn't make a mold of the part using a cast from a 3D print, as long as the print was sanded smooth before doing so.
As for responsibility when it fails, it's dependent on how many test prints were run and if the user was careful with the piece. Prints have a tendency to be fragile, especially if the object being printed is not necessarily "self sustainable" upon it's own weight.
A gear, (being a solid block,) is inherently more structurally sound than a leaf, as an example. The leaf is broad in height and width, but is paper thin and will tend to bend, fold or tear, three things not commonly stated about gears.
At this point in time, 3D printing is still mostly for novelty or aesthetics, or light duty mechanicals if any mechanicals at all.
3D printing as a concept is still in the experimental stage.
I think his sunroof gear qualifies as light duty.
I'd say CNC is the way to go if durability is part of the question, or if there is significant heat near where the part is going.
I hope this answers a few questions... ?