The Report Card with Nat Malkus
Sep 23, 2026 · 1 hr 8 min · 14 segments
Proponents of education technology have often made big promises about its potential to transform schooling, but so far, ed tech has failed to live up to its promise. Will AI also fail to deliver, or…
Philip OreopoulosGuest
Nat MalkusHost
As a middle school teacher, a former middle school teacher, I understand this deeply.
In Conmigo, they've designed the AI, the Conmigo, to not just give them the answer, but to try and push them through the material.
How much upside do you think there is for AI-assisted tutoring beyond just computer learning systems or computer-assisted learning programs, if we can get the application right and the engagement up? Is the upside
huge? So that is the key question that a lot of us are trying to understand.
I think that it's a really open-ended question and there's a lot of uncertainty about how far this can go, whether it can actually be a game changer.
I think the way I think about it is, Could these models get close to the experience that we might expect to see if we imagine what a real human tutor would be doing if they were looking over an individual's shoulder whenever they were working on a problem? I'm more optimistic than some of my other colleagues around the potential.
Well, I think there's a lot of more research and a lot of things that we can keep trying to see what would happen if we get closer and closer to that experience, where, for example, the tutor starts to speak, going voice to voice rather than text to text, or starts to understand the student's learning style, incorporates the student's Hobbies and things that they like into some of the tutoring also works on their motivation and tries to understand where that's coming from.
I think there's a lot more things that we can try before we sort of understand what the full potential of this is.
I'm assuming that we should not sort of jump on the implication that this is for remedial learning, but that that was just the context of this experiment.
And so the potential for AI tutors is not just catching kids up, but it's learning sort of across the spectrum, again, if we can get it right.
I mean, my research goes where the administration is willing to let me go.
I think there is the interest and curiosity of the potential for these models to help.
some students who are smart to get even more ahead and work ahead that they wouldn't get in class.
So I think there's the potential for these tutors to help at all levels, and there's no reason why it just has to
As a middle school teacher, a former middle school teacher, I understand this deeply.
In Conmigo, they've designed the AI, the Conmigo, to not just give them the answer, but to try and push them through the material.
How much upside do you think there is for AI-assisted tutoring beyond just computer learning systems or computer-assisted learning programs, if we can get the application right and the engagement up? Is the upside
huge? So that is the key question that a lot of us are trying to understand.
I think that it's a really open-ended question and there's a lot of uncertainty about how far this can go, whether it can actually be a game changer.
I think the way I think about it is, Could these models get close to the experience that we might expect to see if we imagine what a real human tutor would be doing if they were looking over an individual's shoulder whenever they were working on a problem? I'm more optimistic than some of my other colleagues around the potential.
Well, I think there's a lot of more research and a lot of things that we can keep trying to see what would happen if we get closer and closer to that experience, where, for example, the tutor starts to speak, going voice to voice rather than text to text, or starts to understand the student's learning style, incorporates the student's Hobbies and things that they like into some of the tutoring also works on their motivation and tries to understand where that's coming from.
I think there's a lot more things that we can try before we sort of understand what the full potential of this is.
I'm assuming that we should not sort of jump on the implication that this is for remedial learning, but that that was just the context of this experiment.
And so the potential for AI tutors is not just catching kids up, but it's learning sort of across the spectrum, again, if we can get it right.
I mean, my research goes where the administration is willing to let me go.
I think there is the interest and curiosity of the potential for these models to help.
some students who are smart to get even more ahead and work ahead that they wouldn't get in class.
So I think there's the potential for these tutors to help at all levels, and there's no reason why it just has to
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
