Uppercase is objectively easier to read.
The exception is when lowercase is TWICE THE LEGIBLE SIZE, it can be easier to read due to "word shape" - the pointy bits on "d", "h", and "p" make "xxxd" and "xhxpx" distinguishable. REGARDLESS OF SIZE, codes like "1lmnmmn" do not have word shape.
Lowercase also suffers from similar characters like "1l" and "mnw" and others depending on the font. Uppercase also suffers this with O and 0 which is why the cross-bar-zero is used on console fonts.
Base32 took a whole different approach to this second problem by using all 26 UPPERCASE letters, and only 6/10 numerals: 2, 3, 4, 5, 6, and 7.
Early computer engineers made good choices for good reasons - legibility increases accuracy.
