How might a computer-generated IQ test present unique challenges not encountered with traditional assessment techniques?