Wrong. The pixels of the iPad 2 are four times the size of the iPad (4th-gen)'s, in terms of area. Twice the width and twice the height means that they are four times larger.
Conversely, the pixels of the iPad (4th-gen) are one quarter the size of the iPad 2's, in terms of area. One half the width and one half the height means that they are one quarter as large.
Net result: iPad (4th-gen) has 4 times the number of pixels that the iPad 2 has.
The "2x" resolution (264 ppi vs. 132 ppi) describes the linear resolution of the screen, not the surface area per pixel.
As shaky as AI writing often is, I really don't see this particular thing as a problem to a reasonable degree. They didn't specify linear or area dimension, so it's not necessarily wrong. The dimensioning on the images should be enough to infer that they mean linear dimensions. Pixels are almost never specified by area anyway, they're far more often measured by pitch, these days more often specified as pixels per inch, both linear dimension.