25% of the sessions at the biggest edtech conference in the US were about AI. How does that compare to the math, English, science, and administrator conferences?
I truly value your perspective on the overblown promise and potential of AI! THANK YOU! I went from teaching, to working for an ed-tech company, to nonprofit curriculum support and consulting company, to back to a school as an instructional leader. Looking at linkedin and seeing the comments of my former colleagues about edtech, I just shake my head. I truly wonder if they are connected to reality. There are so many kids and teachers in buildings who need the support of PEOPLE - not more computer programs. And I even use AI to do simple tasks for myself and teachers; chat GPT wrote a great article about midwives in the medieval period for our 7th graders reading The Midwife's Apprentice. Our literacy intervention coordinator uses it to write simple passages for fluency practice. Teachers write emails with it. But actual teaching and learning comes from the work that people do together.
I'm reading this building an 'edtech' solution (hate that word). Thank you for curating all this great content. Assuming children learn like adults and school can be treated like a Udemy course is mistake number one.
That said, I see the problem to solve is not automating teaching but to support the environment where learning happens (aka a classroom). This means automating the boring, irritating, and menial stuff that keeps teachers up until 10 at night preparing for the next day. But it also means capturing data (and doing something with it) so there are less children left behind because they don't speak up or don't show up in as failing in assessment.
Great article! Thank you for pulling all this together. Parent perspective: From my experience interacting with young kids (preschool and early elementary) in the past couple of years, I believe that embodied human interaction is crucial in teaching. I feel like technologists have an idealized view of AI enabling accelerated learning for our children, which contrasts with the growing body of evidence around increased distraction, addiction, etc. behaviors that are associated with use of screens by children and youth. I do think that some applications of the learning chatbots or virtual personalized tutors will be helpful - but we’re still in the early days of figuring out the basics of managing screen time with our kids.
Hopefully openai fails to raise funds and then at least two good things will happen: 1. We can be done with this overhyped kabuki theatre of genai. 2. There will be at least one new interesting docuseries to watch on netflex or hbo in a year or two.
Thanks for reporting on what is really happening on the ground.
What is tricky about the change from non-profit to for-profit for OpenAI is that much of the content it was initially trained on was shared with a "non-profit". I am not 100% certain that all of the content would have been shared with a for-profit and I am not sure models built on it can just be handed over to a for-profit. It will be interesting to see if this works and would provide a cheap and easy route to get your hands on lots of data with no cost and then you could switch your model.
Desmos is better than those old plastic jobs. I stuck with those plastic jobs until ladt year because it was the best tool SAT allowed. SAT has made the desmos ui it's on screen calculator, so last year I was finally able to move to desmos without reservation.
Without defining where we see AI benefitting a teachers work, the companies will direct the attention to wherever the most revenue can be generated. And that is usually a program that individual students are using as often as possible. The most important thing school district leaders can do right now is have the courage to pause purchasing or accepting free “pilot” programs so teachers can help define ai’s role in their work rather than others defining it for them. Slowing down when others are moving fast is tough politically, but usually results in a better system.
Suppose the teachers face rotten incentives, or aren't allowed to impose discipline, or aren't being asked to teach the right things in the first place? We can't infer that AI tutoring is ineffective as a technology just because it might not fit well incrementally into current conventional school classrooms, right?
Maybe, but this logic is non-falsifiable. "My magic beans WOULD be effective but teachers have poor incentives to use it and the system is too conventional."
I truly value your perspective on the overblown promise and potential of AI! THANK YOU! I went from teaching, to working for an ed-tech company, to nonprofit curriculum support and consulting company, to back to a school as an instructional leader. Looking at linkedin and seeing the comments of my former colleagues about edtech, I just shake my head. I truly wonder if they are connected to reality. There are so many kids and teachers in buildings who need the support of PEOPLE - not more computer programs. And I even use AI to do simple tasks for myself and teachers; chat GPT wrote a great article about midwives in the medieval period for our 7th graders reading The Midwife's Apprentice. Our literacy intervention coordinator uses it to write simple passages for fluency practice. Teachers write emails with it. But actual teaching and learning comes from the work that people do together.
I'm reading this building an 'edtech' solution (hate that word). Thank you for curating all this great content. Assuming children learn like adults and school can be treated like a Udemy course is mistake number one.
That said, I see the problem to solve is not automating teaching but to support the environment where learning happens (aka a classroom). This means automating the boring, irritating, and menial stuff that keeps teachers up until 10 at night preparing for the next day. But it also means capturing data (and doing something with it) so there are less children left behind because they don't speak up or don't show up in as failing in assessment.
Great article! Thank you for pulling all this together. Parent perspective: From my experience interacting with young kids (preschool and early elementary) in the past couple of years, I believe that embodied human interaction is crucial in teaching. I feel like technologists have an idealized view of AI enabling accelerated learning for our children, which contrasts with the growing body of evidence around increased distraction, addiction, etc. behaviors that are associated with use of screens by children and youth. I do think that some applications of the learning chatbots or virtual personalized tutors will be helpful - but we’re still in the early days of figuring out the basics of managing screen time with our kids.
Hopefully openai fails to raise funds and then at least two good things will happen: 1. We can be done with this overhyped kabuki theatre of genai. 2. There will be at least one new interesting docuseries to watch on netflex or hbo in a year or two.
Thanks for reporting on what is really happening on the ground.
What is tricky about the change from non-profit to for-profit for OpenAI is that much of the content it was initially trained on was shared with a "non-profit". I am not 100% certain that all of the content would have been shared with a for-profit and I am not sure models built on it can just be handed over to a for-profit. It will be interesting to see if this works and would provide a cheap and easy route to get your hands on lots of data with no cost and then you could switch your model.
Your reporting is a bright light in an otherwise very dark world.
Desmos is better than those old plastic jobs. I stuck with those plastic jobs until ladt year because it was the best tool SAT allowed. SAT has made the desmos ui it's on screen calculator, so last year I was finally able to move to desmos without reservation.
Without defining where we see AI benefitting a teachers work, the companies will direct the attention to wherever the most revenue can be generated. And that is usually a program that individual students are using as often as possible. The most important thing school district leaders can do right now is have the courage to pause purchasing or accepting free “pilot” programs so teachers can help define ai’s role in their work rather than others defining it for them. Slowing down when others are moving fast is tough politically, but usually results in a better system.
"Slowing down when others are moving fast is tough politically, but usually results in a better system."
Extremely on point.
Suppose the teachers face rotten incentives, or aren't allowed to impose discipline, or aren't being asked to teach the right things in the first place? We can't infer that AI tutoring is ineffective as a technology just because it might not fit well incrementally into current conventional school classrooms, right?
Maybe, but this logic is non-falsifiable. "My magic beans WOULD be effective but teachers have poor incentives to use it and the system is too conventional."