The human-targeted text in this design is using features with a low spatial frequency, compared to the robot-targeted text at a higher spatial frequency. Squinting blurs your vision so the high frequency details are lost.

Thanks for explaining, that does make sense