The study itself acknowledged this, saying, "a meta-analysis showed that greater screen use was associated with decreased language skills, whereas screen time spent on educational programs was associated with increased language skills," and, "A limitation is that the information we collected did not allow us to separate educational screen time from other types of screen time."
You're right that babies won't be able to engage with games or puzzles for 4+ hours per day, but that is not what others in this thread are suggesting. The criticism I'm seeing is that a parent interested in educating their children shouldn't throw out all screen time just because most other parents use screen time poorly.