The writer focuses on the education of children, in a school context. He does jump around a bit in age group - young children, late teenagers, college students. He is persuasive that young children /in a school context/ are best educated by one or more adult teachers, in small groups.

But then, the only alternative in that context would be the same but with a higher pupil:teacher ratio (at the limit that might mean zero physically-present teachers, like the tv example he gives). And yes, that will generally be worse, holding the student, the peers, and other variables constant.

One case he omits is a 1:1 teacher:student ratio, perhaps because that isn't school. It's tutoring.

The results of personalized 1:1 tutoring are actually a lot better than mass education according to the quasi-industrial institution of 'school'. This is well-studied, I don't think it is controversial among educationalists.

The same goes for other approaches to education, whether of children or adults.

School is actually a fairly historically specific phenomenon, and it may not be the best approach we can take (even as a system for compulsory mass education of children to legally prescribed minimum standards, bundled with free daycare).

If technologies generally fail to improve school-based education, perhaps that says more about the inherent limitations of that institutional framework than it does about the scope of those technologies. Perhaps the right technology might unlock better results: for self-education, for students paired with individual tutors, and for whichever other models might exist. But the actual model of education might need to be different from traditional 'school'.

To paraphrase Henry Ford, the education system will only consider ways to develop a 'faster horse', rather than embracing approaches that might replace equine transportation altogether. Fair enough, but we don't have to share their assumptions.