I feel like learning to steer it would be just as possible as learning to ride a bike you steer and pedal in reverse. You just need to get past your muscle memory.
Unless I'm misunderstanding the physics, it would be quite literally impossible, not a matter of learning - any attemp to steer results in the angle of the bike moving in the opposite direction an amount directly proportional to the change in direction, and with gravity being cancelled out by the spring (or in 0G), there is no way to rectify the tilt except by steering back to the original course.
Technically, you can steer temporarily - if steering right by 20° tilts you to 45°, you can ride at 45° (actually, it's possible that a 45° turn would always relate to a 45° tilt - I'd need to think on it), however your allowed directions are determined entirely by your starting direction; if you started pointing South, there would be no possible way to move North, and would likely only be able to operate the bike in the 160° - 200° range, looking at the video.
Mathematically, with the usual simplifications and restrictions we mean by that, the device can not be steered freely at all.
Practically, if you were left in a room with the device alone for a while, you'd probably find some way to change its direction. My initial guess is to jerk it up as high as possible and then jerk it sideways and rebalance. It's pretty heavy, so this isn't going to work very well and will be very exhausting. However, I suspect the ultimate technique you'd settle on would work just as well if we locked the front wheel entirely so it couldn't be turned, so while you may be able to "direct" the device you would still arguably not be "steering" it.
Heck, you might turn slowly if you just rode around tilted, due to the way the tires deform - although I've nowhere near enough familiarity with tire physics to know which direction.
Yup. There's probably a lot of little such things where the model deviates from reality. Might be a way to exploit the friction in the various hinges, etc. It may be easier to turn on a low-friction surface where the ground friction doesn't overwhelm these effects, etc. Fun to play with, but the end result is still that even if the real world doesn't give us a "truly" unsteerable bike, I am still surprised that this curve between "steerable bike" and "steerable trike" passes near "zero steerability" at all. I would not have guessed that.
The biggest problem with turning in zero gravity is there's no other force than forward momentum.
The only thing that keeps bicycles upright in gravity is forward momentum. If a bicycle in gravity with no rider tried to turn right on its own, it would fall over. Only if the road curves naturally to the right will it follow and not fall over.
Imagine no bike. You're floating forward in zero-G. Now you want to go right. You turn to the right.... but you're still going forward. It's like that, but on the bike. Nothing is keeping you stuck to the ground, so no matter how you try to move the bike, it will want to keep going in the original direction of momentum, and you will just end up tumbling over if you try to turn the wheel or lean anywhere.
Another way to look at it, just like in the video: if you are tilted, you keep going forward, while tilted. You would literally need some force to pull you in a direction other than forward in order to turn without tumbling toward your original direction.
With a tricycle, when you turn the wheel, the inside wheel is essentially anchored to the ground where it is, and the outside wheel follows the only path that it can, since it can no longer continue going forward. If you were going fast enough during this turn, the whole thing would tip over, similar to how cars in gravity will flip over when they try to turn too fast. Momentum just carries them forward.
