It is a case of denying the antecedent
Let's get the informalities over first.
The argument is invalid because it tells us nothing about how inexperienced driving and irrationality are related. Specifically it does not tell us that all and only inexperienced drivers are irrational. Since the argument gives us no reason to suppose that only inexperienced drivers are irrational, it leaves open that other groups are also irrational and that Harry belongs to one or more such groups.
Denying the antecedent
p → q
In other words :
For all x, if x is an inexperienced driver then x is irrational.
x [Harry] is not an inexperienced driver.
x is not-irrational
What's wrong here is that in the second premise the antecedent of the first premise [x is an inexperienced driver] is denied instead of the consequent [x is irrational].
Put it this way. In a deductively valid argument the conclusion cannot be false if the premises are true. But here the premises 1. & 2., could both be true and yet the conclusion, 3., false.
It might indeed be true that 'For all x, if x is an inexperienced driver then x is irrational' and 'x [Harry] is not an inexperienced driver' but the conclusion, 'x is not-irrational' could still be false because x, here = Harry, could be irrational by virtue of his being psychotic or neurotic - a condition which has nothing to do with his being a driver, experienced or otherwise.
The original question scouted the possibility of undistributed middle. Let's dispose of that.
In a very compressed nutshell, the middle term is the term which appears in both premises but not in the conclusion. Here the middle term is 'inexperienced driver' and it is properly 'distributed'; it occurs in both premises and is absent from the conclusion. So the middle term is distributed and this is not a case of undistributed middle.