I get that this is just a meme, but for those who are curious about an actual mathematical argument, it is because Pythagoras's theorem only works in Euclidean geometries (see proof below). In Euclidean geometry, distances must be real numbers of at least 0.
There exists at least one ∆ABC in a 2-D non-Euclidean plane G where (AB)² + (AC)² ≠ (BC)² and m∠A = π/2
Proof: Let G be a plane of constant positive curvature, i.e. analogous to the exterior surface of a sphere. Let A be any point in G and A' the point of the furthest possible distance from A. A' exists because the area of G is finite. Construct any line (i.e. form a circle on the surface of the "sphere") connecting A and A'. Let this line be AA'. Then, construct another line connecting A and A' perpendicular to the first line at point A. Let this line be (AA')' Mark the midpoints between A and A' on this (AA')' as B and B'. Finally, construct a line connecting B and B' that bisects both AA' and (AA')'. Let this line be BB'. Mark the intersection points between BB' and AA' as C and C'. Now consider the triangle formed at ∆ABC. The measure of ∠A in this triangle is a right angle. The length of all legs of this triangle are, by construction, half the distance between A and A', i.e. half the maximum distance between two points on G. Thus, AB = AC = BC. Let us define the measure of AB to be 1. Thus, 1² + 1² = 2 ≠ 1². Q.E.D.
Read the "1" unit side as "move left 1 unit" and the "i" side as "move up i units", and the hypotrnuse is the net distance travelled.
The imaginary line is perpendicular to the real line, so "up i unit" is equivalent to "right 1 unit". The two movements cancel out giving a net distance of zero.
Yep. A vertical line segment above A with length 𝑖 is a horizontal line segment to the left that's 1 unit long. So, the diagram needs a "not to scale" caveat like a map projection, but there's nothing actually wrong with it, and the triangle's BC side is 0 units long.