While I am sure Alpha Schools might “work” for many students, just as many models of education “work” for many students, I am curious about the long term impact of this model in particular. The term, “going postal” was coined in the 1980’s following a series of violent incidents involving postal workers. The postal office culture of rigid hierarchies, strict management practices, high pressure, and isolated work environments helped create this phenomenon. For Seinfeld fans out there, Newman is great example of how impactful “it never stops” as a model is. So when a “school model” is funded by a company that engages in unscrupulous business practices, such as exploiting workers using heavy surveillance, it is challenging for me to believe that their school model is for any other purpose than to produce more workers and fewer thinkers. As well, what works for one child with autism will cause another child with autism great distress. I respect parenting experience - and from my own experience both personal and professional having worked with hundreds of students with specialized learning needs including autism, I must push back on this idea that this environment will help them feel like they are not the “bad kid”. And finally, as a teacher who taught at a “DI School” (ha - before “AI”, there was “DI”) this scripted instruction did not work for my students with special learning needs at all. All of my 4th and 5th graders were placed in the lowest level (Decoding A) with zero nuance which meant I had kids who didn’t know the alphabet mixed with kids who had a little bit of reading skill under their belts. When following the script, I had to snap my fingers at the children (to pace their spelling) and if any child in my group of 8 special needs students made a mistake, I had to start the entire lesson over…I can still hear me saying, “starting over” and then getting a chair thrown at me as the student tired of hearing this. I guess the good news here is that when said student throws said chair at AI, it won’t bleed out as a human would. Silver lining.
The Bjorks' desirable difficulties framing is underrated, and the implication for AI tutors is uncomfortable. If struggle is the mechanism, an AI that's too smooth, too fast to fill the gap, might be quietly undermining the thing it's supposed to build. The question worth asking: what does "personalized" mean when the optimization target is engagement rather than encoding difficulty?
I tried to find peer-reviewed research by Prof Hendrick but couldn't find any, apart from a rather dense thesis. Know any? I don't mean repeating other people's work like in this interview.
While I am sure Alpha Schools might “work” for many students, just as many models of education “work” for many students, I am curious about the long term impact of this model in particular. The term, “going postal” was coined in the 1980’s following a series of violent incidents involving postal workers. The postal office culture of rigid hierarchies, strict management practices, high pressure, and isolated work environments helped create this phenomenon. For Seinfeld fans out there, Newman is great example of how impactful “it never stops” as a model is. So when a “school model” is funded by a company that engages in unscrupulous business practices, such as exploiting workers using heavy surveillance, it is challenging for me to believe that their school model is for any other purpose than to produce more workers and fewer thinkers. As well, what works for one child with autism will cause another child with autism great distress. I respect parenting experience - and from my own experience both personal and professional having worked with hundreds of students with specialized learning needs including autism, I must push back on this idea that this environment will help them feel like they are not the “bad kid”. And finally, as a teacher who taught at a “DI School” (ha - before “AI”, there was “DI”) this scripted instruction did not work for my students with special learning needs at all. All of my 4th and 5th graders were placed in the lowest level (Decoding A) with zero nuance which meant I had kids who didn’t know the alphabet mixed with kids who had a little bit of reading skill under their belts. When following the script, I had to snap my fingers at the children (to pace their spelling) and if any child in my group of 8 special needs students made a mistake, I had to start the entire lesson over…I can still hear me saying, “starting over” and then getting a chair thrown at me as the student tired of hearing this. I guess the good news here is that when said student throws said chair at AI, it won’t bleed out as a human would. Silver lining.
The Bjorks' desirable difficulties framing is underrated, and the implication for AI tutors is uncomfortable. If struggle is the mechanism, an AI that's too smooth, too fast to fill the gap, might be quietly undermining the thing it's supposed to build. The question worth asking: what does "personalized" mean when the optimization target is engagement rather than encoding difficulty?
I tried to find peer-reviewed research by Prof Hendrick but couldn't find any, apart from a rather dense thesis. Know any? I don't mean repeating other people's work like in this interview.