Week in Edtech 8/12/26: Educator AI Adoption Hits 80%, Varsity Tutors Exits Schools, Higher Ed Rethinks Grading, Open vs. Closed AI, and More! Feat. Reg Leichty of Foresight Law + Policy & Mike Wales and Madison McCormick of CodePath

Join hosts Ben Kornell and Alex Sarlin as they break down the biggest stories shaping educator confidence, tutoring, higher ed grading, AI models, education policy, and AI workforce development.

✨ Episode Highlights:
[00:02:47] Edtech Insiders heads to Teach For America’s 35th anniversary
[00:05:19] HMH survey shows educator AI adoption reaches 80%
[00:08:45] Varsity Tutors exits school-based tutoring
[00:11:46] What makes tutoring effective?
[00:13:00] MIT drops freshman grades as Harvard tightens grading
[00:15:06] AI reshapes academic integrity and assessment
[00:17:41] AI cheating accelerates competency-based learning
[00:19:52] Open vs. closed AI models heats up
[00:23:53] Could AI become a public utility?
[00:28:11] New benchmarks measure AI tutoring effectiveness

Plus, special guests:
[00:29:40] Reg Leichty, Founding Partner of Foresight Law + Policy, on screen time and AI regulation
[00:47:26] Mike Wales, Senior Director of AI Programs at CodePath, and Madison McCormick, CodePath leader, on AI apprenticeships and workforce development 

😎 Stay updated with Edtech Insiders! 

Follow us on our podcast, newsletter & LinkedIn here.

🎉 Presenting Sponsor/s:

Innovation in preK to gray learning is powered by exceptional people. For over 15 years, EdTech companies of all sizes and stages have trusted HireEducation to find the talent that drives impact. When specific skills and experiences are mission-critical, HireEducation is a partner that delivers. Offering permanent, fractional, and executive recruitment, HireEducation knows the go-to-market talent you need. Learn more at HireEdu.com.

This season of Edtech Insiders is brought to you by Cooley LLP. Cooley is the go-to law firm for education and edtech innovators, offering industry-informed counsel across the 'pre-K to gray' spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education.

Tuck Advisors was founded by entrepreneurs who built and sold their own companies, frustrated by other m and a firms, they created the one they wished they could have hired but couldn't find. One who understands what matters to founders and whose North Star KPI is the percentage of deals closed. If you're thinking of selling your ed tech company or buying one contact Tuck advisors now.

Every year, K-12 districts and higher ed institutions spend over half a trillion dollars—but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, then helps you turn them into personalized outreach—fast. Win the deal before it hits the RFP stage. That’s how top edtech teams stay ahead.

Transcript

Copy

[00:00:00] Alex Sarlin: Innovation in pre-K to grade learning is powered by exceptional people. For over 15 years, Edtech companies of all sizes and stages have trusted higher education to find the talent that drives impact. When specific skills and experiences are mission critical, higher education is a partner that delivers.

Offering permanent, fractional, and executive recruitment, higher education knows the go-to-market talent you need. Learn more at hireedu.com. That's H-I-R-E-E-D-U.com. 

[00:00:33] Ben Kornell: One of the themes that we've really been hitting on all summer is changing in higher ed grading, and it's not a clear direction where you have a group that are going with no grades, so MIT with no freshman grades, the idea being to reduce pressure, competition around grades, whereas Harvard famously has rolled out a forced grading curve where only 20% of grades can be A's.

We're also seeing AI use being prohibited versus AI use being embraced. Some schools, they're turning to oral defenses with an AI agent. You can do all your assignments with AI. They're kind of like, that's the way the world's going to work. We just need you to defend your knowledge through an oral defense

[00:01:20] Alex Sarlin: Welcome to Edtech Insiders, the top podcast covering the education technology industry. From funding rounds to impact to AI developments across early childhood, K-12, higher ed, and work, you'll find it all here at Edtech Insiders. 

[00:01:36] Ben Kornell: Remember to subscribe to the pod, check out our newsletter, and also our event calendar.

And to go deeper, check out Edtech Insiders Plus, where you can get premium content, access to our WhatsApp channel, early access to events, and back-channel insights from Alex and Ben. Hope you enjoy today's pod.

Edtech Insider listeners, we're back with another Week in Edtech. It's Ben and Alex here, coming to you from our favorite co-working spaces from around the world. Alex, so glad to have you here today. What's going on in the pod? What's new in the, in pod land? 

[00:02:16] Alex Sarlin: Yeah, we got some really fun conversations coming up.

We just published one with Boddle, with Edna Martinson from Boddle Learning, who is, was amazing. We talked to Student Achievement Partners, Betsy, and Tony from EdSurge. I think soon we're-- the coming up in just a couple of weeks is Aneesh Sohoni, I may be mispronouncing that, who's the CEO of Teach For America, which is gonna be amazing.

We also just announced, Ben, right, that we are going to be on stage at the Teach For America thirty-fifth anniversary. You wanna, you wanna break the news here? 

[00:02:45] Ben Kornell: Uh, no, you go for it. 

[00:02:47] Alex Sarlin: It's really exciting. I- you made this happen. We are going to be on stage in front of somewhere in the neighborhood of seven to ten thousand teachers at the TFA thirty-fifth anniversary in Las Vegas in just a couple of months, interviewing Lin-Manuel Miranda, who is a, an ex-teacher, an ex-substitute teacher, somebody who has been in education on and off for quite a long time, and has all sorts of thoughts about education and creative education.

It's gonna be a thrill. 

[00:03:13] Ben Kornell: We're not gonna throw away our shot, that's for sure. 

[00:03:16] Alex Sarlin: Yeah. It's gonna be a blast. He's gonna be a s- a real headliner. I think we're opening the conference, is that right, Ben? 

[00:03:22] Ben Kornell: There's the opening remarks, and then we're first up on the agenda. So it'll be a, a really fun time, and also a chance to reflect on the importance of arts and arts education, the, the impact that, you know, singular teachers can have on the trajectory of, of kids, and also Lin's work in Puerto Rico and with arts education across the country.

It's going to be a great event. If you're a Teach For America alumni, alumnus, make sure you sign up. If you're not, they do have a vendor expo, and they are open to Edtech companies coming and, and sharing more about what they're working on. We'll have a booth there for EduHam, which is actually the curriculum for social studies that people can use based on Hamilton.

So it'll be a great time. We also, for today, we have Mike Wales and Madison McCormick from CodePath, and Reg Leichty from Foresight Law & Policy. Reg is really an expert in the policy changes going on in our landscape, especially around screen time. So everything from big name hit Broadway musical stars to, like, deep diving into your latest political expertise, we've got it all here at Edtech Insight.

[00:04:38] Alex Sarlin: And don't downplay the CodePath. CodePath is the official vendor working with Anthropic for Claude Core, where they're, they're, uh, teaching all of these people to be nonprofit apprentices and basically doing all sorts of really powerful training in association with Anthropic. And CodePath has a, a great record of serving underserved communities in very important workplace skills.

So in- incredible guests. We also have Bellwether coming up, NWEA, Carl Mahdy from StepFull, who we've never talked to. StepFull is a, a really, uh, successful Edtech company we've 

[00:05:11] Ben Kornell: never- Can't believe we had never had them on, yeah ... 

[00:05:13] Alex Sarlin: we've never had them on. So it's gonna be great. And I think there's a happy hour coming up.

Is that right, Ben? 

[00:05:19] Ben Kornell: We also have happy hours coming up. We're gonna announce our full schedule for the fall coming up soon. We'll be hitting San Francisco, Los Angeles, New York City The idea is that we'll be a traveling roadshow to bring the Edtech Insider community together. So stay tuned for that. More details to come.

Speaking of, like, around the world here, we had a great starter survey. It's the 12th annual survey from Houghton Mifflin Harcourt around educator confidence. And the origin story of that was really around the teacher shortage. You know, teachers were leaving in droves. What was driving that? And today, it just dropped the 12th version, and we saw some really interesting gains, more retention, 17% increased, uh, retention of administrators, but also a lot about AI.

So over the last three years, AI adoption has jumped from 10% of teachers using it regularly to 80% in just three years. But AI confidence is flat to down, with only two-thirds feeling confident in how to use AI. And there's some dissonance. 87% feel that their AI use is instructionally effective, yet only 62% know what to teach students for an AI world.

So really interesting survey that just shows adoption and penetration really outpacing the understanding of what efficacious use of AI is. 

[00:06:48] Alex Sarlin: Absolutely. And you know, a couple of other findings. We've talked to Francie Alexander from HMH. She's actually an old friend. We've talked to her a couple times in the past.

Maybe we can bring her back on to talk through these findings 'cause they're really-- there's some interesting ones here. I mean, as you mentioned, confidence in the profession is up. You, we, you saw a big jump in educators saying that they have no interest in leaving education. As of even last year, that was only a third of educators had no interest, meaning two-thirds have not ruled out leaving education.

They were sort of looking at, at least looking at other options. That's gone up to 50%, 52% in one year. So from a third of educators sort of feeling, "Hey, I'm here and I'm staying," to half. That's a big jump. 

[00:07:29] Ben Kornell: I mean, the, the ru- the grass is not looking as green on the other side, I think. Part of it may be that there's increasing satisfaction with the role, and it might also be that When you're looking at the lack of dependability of these other career moves, I think teaching is looking like a much more attractive space to be in.

[00:07:50] Alex Sarlin: Exactly. But at the same time, 70% of educators are not sure that education policy is going to sort of solve itself, is going to work through, is going to drive progress or improve teaching and learning. So there's not more confidence that the profession itself and the administration and sort of the thinking that goes into the, the education world is getting, you know, is, is getting more sophisticated and the policy is catching up.

And we've seen lots of teacher findings that the policy is, is way behind, especially when it comes to AI. And they're still-- they still don't think it's gonna catch up. So it's an interesting sort of conflicting findings. And as you said, 10% were using AI in 2023, 50% in 2024, 70%, almost 70 in 2025, and we're at 80%.

I remember Dan Meyer said something a while back. I had a prediction in one of the year-end episodes that every teacher is gonna be using AI in a year, and he's like, "That's crazy." 80% is pretty high. It's pretty high. Vast majority. 

[00:08:45] Ben Kornell: One other thing that's been interesting is in the kind of AI is going to take everybody's job versus no one's going to use AI at all.

Teaching has been one of the professions that really shows durability in an AI world, and I think that's also been showing in the tutoring space. So we actually had a really interesting story around Varsity Tutors, which is a B2B company only as a secondary business. It really started as a B2C business, but they are shutting down their tutoring service for schools.

And I, I wanna make sure people understand this was a multimillion-dollar business. It's not like they were a tiny little business. But what I think they found is that focusing on their core B2C was a much steadier growth path, but they also saw lack of evidence of impact in their tutoring with schools.

So I think it's, you know, forcing a lot of people in the space to ask the question Is tutoring in the school environment dead on arrival, or is it just that tutoring programs like Varsity Tutors don't work and other ones will work? I'm curious what your take is given all the fallout, and we've been following this both on mainstream media and then on our back channels.

What's your take on the Varsity Tutor news? 

[00:10:05] Alex Sarlin: We've talked to the-- Chuck Cohn from Varsity Tutors. They've had a business for a long time sort of selling to individuals, adults, selling to parents, selling directly to consumers for all sorts of different type of tutoring. My assumption here when I look at the news but also sort of put together the dots here, I think it's a combination of a few different things, right?

I mean, funding has continued to shrink, obviously, since ESSER. I think the sort of glow around tutoring as a solution, you know, if we talked five years ago, people were like, "Tutoring is the gold standard. Tutoring is the thing that works. We all know that throwing money towards tutoring is, like, a sure thing."

And then that-- the story has really changed in the last few years due to a number of big stories, mostly because of the dosage issue, and you're seeing that come up in the news here too. There's just this idea of, like, be able to implement a tutoring program. It's not a matter of giving people access.

Access is just the very beginning. It's actually a, a whole complex structure to actually get all the different components of the tutoring equation, the tutors, the students, the, the administrators within the districts to, uh, principals, school teachers, a-after school coordinators, like, to all match up to make it actually work consistently.

And I think what you're seeing is just, you know, with-without enough dosage, you don't get the impact. Without enough impact, you don't see any movement, and schools say, "I don't see this working." And the blame can go to any par-- you know, par-- you can sort of point the finger towards anybody in that equation.

But I think that Varsity, after doing a lot of individual tutoring, was just not seeing all the pieces click into place. And given the decrease in funding, probably said, "It's not worth it for us to build up this business and throw a ton of money and people at the implementation solution." That's my take.

[00:11:46] Ben Kornell: Yeah. You need to have operational integration with schools to get to the dosage for sure. I think the other two things that have proven important is relational connectivity. And so we're seeing success with programs where it's like a college student or a near-peer tutoring run by a nonprofit because they're building a relationship over time with the students rather than kind of substituting in and out a tutor.

And then I think the third one is integration with the curricular program at the schools. When tutoring is happening in a totally separate vector where how you're being taught math is happening one way over here and how it's being taught in another way over here, it actually adds to the confusion. So we're seeing people that have figured out how to more deeply align or integrate that that's really working.

I think the other thing is that Varsity Tutors getting squeezed by nonprofit volunteer tutoring programs and pure AI play tutoring programs of third space learning, which is coming into the US with a one hundred percent AI driven tutoring, which is essentially unlimited dosage. And so, you know, a model like theirs fits better with the B2C motion I know we've been talking a lot about K-12.

Let's talk a little bit about higher ed. So lots going on in higher ed. We're really excited. We've got a brand new series that we're launching at the end of this month with Matthew Rascoff from Stanford, and we're gonna be talking to some of the big names in higher education. Our series kicks off with the president of Oklahoma University, the president of Southern New Hampshire University, and David Rogier, the CEO of Masterclass, starting a conversation about the future of higher ed.

But one of the themes that we've really been hitting on all summer is changing in higher ed grading, and it's not a clear direction. It's kind of polarities where you have a group that are going with no grades, so MIT with no freshman grades. There's a bunch of other schools that are starting to follow suit with no freshman grades.

The idea being to reduce pressure, competition around grades. Whereas Harvard famously has rolled out a forced grading curve where only twenty percent of grades can be A's. We're also seeing AI use being prohibited versus AI use being embraced. Some schools, they're turning to oral defenses with an AI agent, so you can, you can do all your assignments with AI.

They're kind of like, "That's the way the world's gonna work. We just need you to defend your knowledge through an oral defense." Actually, the-- there's a company in Australia that we just interviewed for the pod that's moving that work forward. And then we have the Con TED Institute, which was just announced two weeks ago.

Sal joined our podcast yesterday, and he really talked about moving from this discrete grades, this virtue signaling from a brand standpoint, and moving much more into concrete portfolios of work where it can be assessed for competency aligned to jobs and careers. So a lot going on there. Those are just a few that I saw.

What did you see in this space? 

[00:15:06] Alex Sarlin: That's a lot of it. I think there's, there's three sort of through lines here that I see that are all interweaving together. One is obviously we've talked forever about the integrity issue that, that has sort of entered the higher ed space because of AI. We saw this week Blackboard acquiring a authorship-first writing integrity startup called Cursive Technology.

We see companies like Turnitin trying to do this sort of gated, gated approach to make sure that they can monitor student cheating. That has thrown the entire sort of concept of grades up in the air, and as you say, people are moving to oral defenses, to blue books, to much more in-person, you know, just a, a whole different thing.

The, the, the need for in-person examinations has also had a big effect on online learning. There was an interesting article, I think, in The Times this week about how a lot of online courses-- Because at this point, you can actually get an agent to basically complete an online course for you. It really throws it into disarray because it's not an option to do an oral defense.

It's not an option to have people sit in a room and do a blue book You could do Pearson vue. There are some versions of this, but for the most online schools, you can't do this. So this, this whole world of academic integrity, Pangram Labs raised $9 million recently to do academic integrity. Then there's this story about what are grades really for in the university setting, right?

In a transactional world where people are paying a ton of money, they want the grades to sort of come with it, right? And people don't wanna spend their whole, you know, life getting into MIT or Harvard or University of Michigan, which also just changed some g- their grading stuff, and then get there and get Cs, and your parents are angry at you, and you're ch- confused, and th- maybe you take a year off.

And I mean, like it's sort of like we-- this weird thing where like nobody wants low grades, including the schools, maybe some of the professors, but including the schools. So the, the grading curve and all of this stuff, this idea of like, how do we actually fight grade inflation? And then the third one, as you mentioned, Ben, is this idea of like, well, in a world where we're moving to AI, in a world where grades aren't-- don't matter as much, what is the future of portfolio-based learning, of mastery-based learning?

And we've seen two interesting moves there that I think are worth calling out, right? We saw the Mastery Transcript Consortium, which had gone to ETS, be resold recently to a company called Legend.org, which is a for-profit sort of AI company, and they're starting to think about what is the sort of portfolio approach.

And we've seen this really interesting change at, at this concept of, of mastery. We've seen that Aaron Rasmussen has acquired the Mastery Transcript Consortium and is putting together the Mastery.org, this whole idea of competency-based transcripts. So you see competency, you see grade inflation, and you see AI cheating, and they're all coming at this concept of what is the point of traditional grading.

And I, I, I think, you know, maybe it's enough to actually make a change 

[00:17:41] Ben Kornell: When you see all of these pieces moving, it feels like still a little bit of a lack of coordination, but it's almost like a market force where now instead of people trying to pioneer and articulate the reason why, we're actually getting momentum towards competency-based, mastery-based learning.

I think the shadow of the cheating is actually a burning platform to move away from discrete knowledge into performance assessment too. Exactly. And so, like, in a weird way, it's actually this, like, war against cheating that could actually bring about the most progressive forms of education that we have.

[00:18:25] Alex Sarlin: Or go back to the oldest way, the oral defense is like one of the oldest things in, in education. Yeah. 

[00:18:31] Ben Kornell: Yeah, and, and, and maybe-- And I'm also not gonna complain if there's a combination of both. We don't need to have it be one or extreme or the other. But digital portfolios as a way of assessing people's work, I think is going to become much more mainstream in hiring practice, because in the hiring process now, you really can't tell what somebody is representing, whether it's true or not, what like-- AI can produce so much of their stuff.

And so actually having a portfolio of your work as part of a job application is making, making way more sense. And because you can instrument the review of that through AI, it's not as big of a labor issue anymore for the hiring company. So it makes sense that K-12 and higher ed would move in this direction.

I just find it ironic that while all that's going on, our earliest university from the 1630s is saying, "We've got to grade on a curve, and only twenty percent of you get an A." That was totally my Harvard undergrad experience. They should have just given us all knives and daggers and said, "Good luck." And what do they say in Hunger Games?

[00:19:40] Alex Sarlin: May the something be with you. May fortune smile upon you. Is that it? 

[00:19:45] Ben Kornell: Yeah. May fortune s-smile upon you. 

[00:19:48] Alex Sarlin: Forever be in your favor. May fortune forever be in your favor, I think. 

[00:19:52] Ben Kornell: May fortune be ever in your favor, Harvard undergrads. I'm so sorry that, that you have to deal with that. Before we wrap up today too, I also think it's important for us to talk a little bit about what's going on in big tech.

There's a big battle going on that we've covered for a long time between open models and closed models, and I think that's reached a fever pitch now, and good old Mark Zuckerberg has thrown his lot into the fire. Unsurprisingly, open models probably benefit Meta disproportionately than others. There's a question of whether they went for open models because it was a belief system or it was actually the only way that they could win given that they were behind the other models.

But he has a huge manifesto on why open models are a good idea, why there shouldn't be control of a powerful few over the models. Meanwhile, the rest of the industry dominated by OpenAI and Anthropic are saying, "This is super dangerous," and they're adding even more safeguarding. One interesting concrete thing is Anthropic plans to add invisible watermarks to AI-generated text.

Interesting for our AI detectors out there. You and I have always been fan-- W-we've been fans of open because we want access to everyone, and for Edtech, having it cheap is, like, really valuable. But there are these issues of safety and data privacy and so on with open models versus closed models, and also the threat of AGI and what it could do to our world.

What's your take on this latest chapter of the open versus closed debate? 

[00:21:27] Alex Sarlin: I think it's worth calling out that, uh, that a lot of the debate about open models right now, and frankly, uh, you know, this moves so fast. I, I-- some of it is beyond my pay grade. I have to do a little more of my deep dive research on the open model stuff.

But you've seen some of these Chinese open models move incredibly quickly, drop costs like crazy. I've seen some, some really amazing price tags on the cost of, uh, of API calls or token calls for, for open source models coming out of China versus the Anthropics and OpenAIs of the world, which I'm sure is getting them very, you know, they're pulling on their collars a little bit and saying, "Okay, wait, what, what is this gonna mean?"

Like, decreased cost is a good thing. Relying on open source Chinese models is probably not a good thing. And when you actually try to sort of play it out over time, I think there are probably very good arguments on both sides for why you don't want, you know, a few mega corporations especially, that are about to be public corporations, that are gonna be responsible for their shareholder values, trying to own the secret sauce and the secret formulas and all the weights, and have the Coca-Cola formula in a safe, right?

Like, I get that. At the same time, I'm sure there's lots of good arguments about why releasing open, really, really powerful open models out in the world is going to enable all kinds of bad actors, allow a race to the bottom in terms of cost, which is a good thing, but also allow a race in the bottom in terms of entrants.

You can have people creating these incredibly powerful AI tools for very little money that can do all sorts of things that I think we have not yet encountered in the world. So it's a little beyond my, my pay grade. I think for Edtech, I think for Edtech itself, because Chinese models and because the sort of fully open source models are, are-- Some of them might not pass muster with any kind of compliance procedure, or there's a little bit of like if you're a magic school or if you're a, you know, an Edtech company and you're trying to figure out what models to use, I don't yet see them going full force to fully open source models.

I think there will be a lot of risk there, but I can imagine them playing with some aspects of open source models and thinking about what they could do there, perhaps for internal stuff, perhaps for content generation that would then have a human in the loop. Like, I can imagine that as people are starting to put together their model costs for different things, having a really safe, secure, you know, model owned by an OpenAI for certain things and open source models for other things is probably where things are actually gonna go.

That's my best guess. 

[00:23:53] Ben Kornell: My take on this is, are AI models heading towards commoditization or are they headed towards differentiation? And if it's headed towards commoditization, the fastest way to get there, if that's the inevitable terminus, is through good competition with open and closed models and kind of the cycle that we're on.

And it feels to me like, from what I can tell, we are on a commoditization path Where like really the delta between the frontier model and the latest open model is a six-month delta, and the capability delta is relatively small. Now, I'm not seeing the top secret ones that are too safe to launch. I think the other challenge that I see is the Chinese models, when you ask them, "What is the name of your model?"

It says, "Claude." So- ... so there's a way in which the closed models are just getting copied, and they're not doing a good job of like protecting their IP, and they need to think more about it, and the US needs to think more about what does IP look like in this space. Funnily, their cries about IP ring a little bit hollow when they copied the entire internet.

So where does a commodity end up going or where does something go when it's commoditized but it is highly risky? That's when it becomes utility. And I think that there's a world where we end up having OpenAI, Anthropic, and maybe two or three other models offered as public utilities, where governmental regulation says, "Here's what's allowed and not allowed, and here's what's paid and not paid."

And sure, you can generate your own electricity in your backyard however you want, but the reliable, safe, and cost-controlled thing is the public utility. And I think that that's a reasonable space to get to, and I think European countries will get there faster than we do. But I wouldn't be surprised if we thought of LLM and that AI layer as a public utility soon.

By the way, that's the best way to even out the workplace arms race is actually just to make public utility AI available to all 

[00:26:14] Alex Sarlin: Fair. I haven't looked into Mistral recently, but that was a, is a European open model creator, and I wonder if that might be one that is actually going to be the one that sort of leads the way in terms of the, your, you know, regulated commodity utility model.

[00:26:30] Ben Kornell: Yeah. And, and you know, to be honest, like how long is it gonna take the UK to have their own proprietary model? I think they're probably not in a hurry right now. DeepMind is located there. But you could imagine that basically each government has their own preferred utility AI that has their own constraints built into it.

And, you know, it's probably like from an entrepreneurial standpoint, you're like, there's some real friction that that will cause, that you, you're probably gonna get a B+ version. But what you just said, which is that actually AI products are a mix of different models, having that as a baseline to say, "Okay, the electricity is powering it.

Now I'm gonna soup up the engine with add a little bit of this model and a little bit of that model," actually makes a lot of sense. And what we've seen is smaller models perform better on specific tasks when trained on specific data. And for Edtech use cases, I think that's compelling too. So we may be in the Wild West here, or AGI recursive AI may just make this a moot question because our AI overlords will decide this for us.

But I could imagine a world where you have running water, you have electricity, you have like a standard AI model that is powered at a very low rate per token that's available to everyone in the US, and then proprietary companies can develop their small-- own small models or borrow open source models to soup it up to make whatever it is they individually or collectively need.

Anything else you wanna throw in from a headline standpoint before we send it off to our guests? 

[00:28:11] Alex Sarlin: Well, I'm gonna write about this and put some of it out, uh, in, in the newsletter soon, so we don't have to beat it to death here. But there's some interesting movement happening on the tutoring benchmark world, which is something we don't talk about that often, but there are a few of these benchmark, basically assessments for models to see how good they are at tutoring in various contexts for various subjects, for various ways.

And you know, this is something that philanthropy has been very interested in because it's sort of, there's been not really a very clear way to determine what models are good for educational purposes, especially tutoring. And some interesting news starting to come out here. People are starting to lean into some of these models.

The, some of the models that are being used to evaluate and compare different, different frontier models and see how they're doing. And I, I'm gonna write it up, but it's, it's an interesting space and it's one that I recommend our, our audience look at, especially if they're doing anything related to tutoring, because it-- we may soon be entering a world where there will be an actual assessment or a, a couple of sort of accepted benchmarks that you will want to run your tool against and get a score against, including TutorBench, including this new ELBench, and there's a really interesting new one called ELCLA.

Kinda like 

[00:29:16] Ben Kornell: OpenCLA. 

[00:29:18] Alex Sarlin: It is. It's, that's on purpose. But what's interesting is it's trying to figure out long-term relationships. It's trying to evaluate-- It basically has data about relationships between tutors and students over a 30-day period, and it can assess whether the tutoring model is, is starting to actually like learn the student and build relational intelligence, which is like, you know, Isabelle Hau's whole thing.

It's r- some really interesting stuff happening in that space. 

[00:29:40] Ben Kornell: Awesome. Well, this has been great conversation today. Another Week in Edtech in the books. We are now going to pass it over to our interviews. If it happens in Edtech, you'll hear about it here on Week in Edtech. Thanks so much. Hello, Edtech Insider listeners.

Today we're joined by Reg Leichty, the founder of Foresight Law and Policy, where he advises education leaders on the legal, policy, and political issues driving digital learning. His practice focuses on education, telecommunications, student privacy, AI, and E-Rate policy, representing clients before Congress, the FCC, and the US Department of Education.

We have so much to learn from you, Reg. Thanks for joining Edtech Insiders today. 

[00:30:22] Reg Leichty: It's great to be with you. Thanks for the invitation, Ben. 

[00:30:24] Ben Kornell: Before we dive in too much on the future of policy, let's just get a current state of affairs. Looking back over the past year, what have been the biggest shifts in public attitudes and policy shaping Edtech, and which are likely to have the longest-lasting impact?

[00:30:40] Reg Leichty: It's a great question. I mean, we l- right, we're living in this incredible moment of public interest in what's happening with Edtech policy and technology policy generally. And I would say, you know, the biggest thing we've observed in terms of public attitudes over the last twelve to eighteen months, as you've said, is what I think was just a simmering kind of family concern about how often kids are on devices, what they might be experiencing on devices, how that has suddenly, right, kind of come to a boil around the school system.

And so we've really seen a lot of thought after the fact now about some of the uses of technology in schools, just what exactly is happening, what's true, what's myth around this public pushback around the amount of time that, that kids and even adults, right, are using technology. And so that's spilling over into all kinds of debates, both at the state and federal level, where we see a lot of broad policymaking, but especially in local school boards and in settings where local school leaders have to make decisions about what they're doing or not with technology for the foreseeable future.

[00:31:51] Ben Kornell: I think in education in the United States at least, we're very used to policies happening at the school board level. Think about liberty moms, think about COVID policy, and so on. But a lot of this is playing out at the state level, and some at the federal level. What does that look like in terms of bills, in terms of legislation?

Are, are you seeing it more as guidance, or is it coming across as full-on restrictions? And how has that evolved? You know, I'm imagining every state's somewhat different. You know, where was it a year and a half ago, and what are you seeing now? 

[00:32:24] Reg Leichty: I think it's taking a lot of different forms. So you mentioned guidance, so one of the easiest things that governmental agencies can do is publish non-regulatory guidance.

And as AI began to emerge as a common technology in schools, we saw a lot of state agencies publish guidance, right? These guidance materials don't have to go through any kind of formal rulemaking process. They tend to be high level and assistive, I think often connecting local schools to other resources that might be helpful and insightful.

But that was just the beginning. You know, as some of the challenges of the technologies, for example, some of the dangerous activities that we saw very young people engaging with AI chatbots around mental health and some of the dangers that emerged, I think we began to see state policymakers begin to think about what legal authority might be needed to better govern some of those practices.

And now we're really seeing at the federal level, I think the culmination of a lot of work around the social media platforms with measures like the Kids' Online Safety Act, with Congress's, you know, kind of conversation with itself about how to extend the Children's Online Privacy Protection Act to cover older childrens.

And we've also seen, right, some specific legislation focused on AI chatbots, some designed to ensure that young users are only experiencing those tools through family portals or family accounts, and some that are more focused on ensuring there's transparency, informing the user that they're engaging with an AI tool, preventing companies from collecting the data of very young children in those AI chatbot settings.

So we've gone along this evolution of guidance at the state level, maybe board policy at the local level towards the emergence of policy in state law, and now we see Congress potentially moving to some action before the end of the year on some of these other measures that are really designed to, what I think is really addressing some of the lowest hanging fruit in that field.

[00:34:47] Ben Kornell: So all of that is totally understandable from a technology safety standpoint, and we've been on a trajectory where social media and the harms have been well understood. Cell phones and distractions have been well understood. But why do you think educational uses have now become steadily in the, in the crosshairs or school use cases have become in the crosshairs?

A lot of, I think, the Edtech community was caught by surprise given that they believed that their technology was really around a learning use case under the supervision of teachers and educators, and that it was essentially a favored use of technology versus these kind of unfavored uses. But now those lines seem to have really blurred.

What's driving 

[00:35:35] Reg Leichty: that? I think it's a few things. You hit one of them on the head, which is the cell phone issue and what I think a lot of families came to see as being a distraction in schools. You know, we saw state legislatures in their twenty twenty-five sessions really react to that quickly, and I think for the reform groups that were pushing those cell phone bans, I think they seized that momentum to take this broader public concern about some technology uses specifically to the school setting.

[00:36:11] Ben Kornell: And screen time overall. Yeah. 

[00:36:13] Reg Leichty: Yeah. And screen time overall. I think schools often, even looking beyond technology, are often placed at the center of public debates. It might be mental health, it might be nutrition, it might be housing, all of the things that might be struggles for families. We have a tendency to put schools at the center of trying to address those societal challenges, and I think technology is part of that broader trend too.

And, you know, there could be some really positive outcomes to that, right? I think the reaction of a lot of school districts has been to say, "Maybe we can do a better job of engaging our families and understanding why and how we use technology to help ensure that children are graduating ready for success in work or further education."

There's an opportunity to really help them understand that technology use in schools is actually one space where it's very closely monitored, you know, unlike a lot of entertainment or other uses outside of a school setting. And then I think we also have with the emergence of AI, which is such a dramatically new thing for so many people, I think we have an opportunity to really help build the capacity of both students and their families in understanding how to use this technology effectively and safely.

So yeah, I think there's just a lot of reasons that these debates become pulled into the education sector, and that has, uh, both positive and negative kind of implications, right, in instances where, you know, policymakers might overreach, right, and prevent the use of technologies that we need, assistive technologies.

As a kid who grew up on a farm in Nebraska, the potential of technologies to deliver rich educational opportunities into smaller settings, those kinds of things. We wanna make sure, obviously, those things aren't swept away. 

[00:38:10] Ben Kornell: Yeah, it's really interesting around how different states take different stances on restrictions.

Some, like you said, are providing guidance, and each school district, local district comes up with their technology use plan as directed by the state, but it's more of a, "We want you to come up with a plan." Other states have actually capped digital instruction at a maximum of forty to sixty minutes per day for students between kindergarten through fifth grade.

I believe Iowa is one of those states. We saw stuff coming out of Tennessee. And what ends up happening is they end up becoming these blanket bans that then also have interesting loopholes for special education, for computer science, for assessment, and all of a sudden it creates confusion on the ground.

What is allowable, what is not? And so I think there's a concern among the Edtech community that as this policy is being shaped, we don't have a seat at the table. It sounds to me like parent groups are primarily the advocates in favor of screen time restrictions. Is that the stakeholder or advocacy group that is most active at the state level?

And what should or could the education technology industry do to have more of a seat at the table in these legislative conversations? 

[00:39:27] Reg Leichty: Really great questions. I do think that the parent groups are the driving force behind a lot of this work, both at the state level but also at the federal level. I think that as, you know, if, if I look to technology companies or their school district partners, I think what's absolutely essential is for companies to really develop strategies for engaging that, those parent communities, and probably working closely in partnership with their school district customers to say, "How can we..."

And I mentioned this earlier, how can we engage parents in a discussion about what the technology is and is not, how it's helping students learn, how it's helping ensure that assessment leads to data that helps teachers then tailor instruction in, in powerful ways Given the breadth, you know, the speed at which, you know, new technologies like AR are unfolding, I think a lot of parents, a lot of families are feeling really unsettled.

And I think you only overcome that by deliberately building trust with them. And I think that begins with a strong education campaign that companies should be part of. And then I think that's a story that the technology providers really need to take to policymakers at the state and federal level to demonstrate good faith around the technologies they're selling, right?

It isn't just about the bottom line, but can really be a differentiator for the student who might be struggling in a particular subject or might need special supports to really be successful. 

[00:41:08] Ben Kornell: Yeah. I mean, from an impact standpoint, the ability for curriculum and learning tools to be adaptive to students, to track data, to create insights, to support teacher professional practice, there's so much that you're leaving on the table if you regress to pure print or physical tools.

I think a challenge that we have is the education industry itself is generally fragmented. There's a few large players of a significant scale, but even those players only have one or two policy people on their entire team, and we're talking about a kind of 50 state, federal, and local school board layered challenge.

Is this an area where we need to create a new industry collective? Are there industry collectives that you've worked with that you think are good channels for this? And generally, how are you advising your clients to navigate this really challenging time that, by the way, on the flip side, if I'm a superintendent and I have like 10 of my Edtech vendors all saying they wanna help me put together a strategic plan around tech use, that also could feel overwhelming.

So how are you advising people to navigate this? 

[00:42:24] Reg Leichty: There certainly are powerful, and I think really effective coalitions at the federal level, some of which do dive into state policy as well. You know, particularly around some of the professional associations that support CTOs like the CoSNs of the world, SETAs of the world at the state level, industry groups like SIA who commit significant resources to education technology specifically.

They create spaces. For example, those three groups host an annual education technology advocacy day in Washington, DC each May, where companies and educators can go to Capitol Hill and talk about Edtech. One thing that I've experienced with Startup and younger technology companies is that they are less able to, of course, hire staff or have less experience in the policy space.

So I, I really work hard to kind of counsel those entities to not be afraid to engage in the policy process. It feels daunting to business people who are not necessarily well-versed in advocacy at first. But I think, you know, what they find when they're encouraged to talk with their members of Congress or important decision-makers at the federal level, uh, what they find is that those people are really hungry for information.

And I think hearing firsthand from the technologists that are developing these tools can be a really helpful resource to policymakers. So the first thing is get past your concern, if you have any, about being a voice. Find your like-minded peers. You know, call on your business associations. Maybe it's a state chamber, maybe it's a national chamber.

Work with them to encourage them to engage in these kinds of conversations, because there are a few things that are more relevant than what's happening in technology right now that I think policymakers need to know about 

[00:44:23] Ben Kornell: All right, last question. We're gonna have to go out on this one. I feel like we could talk about this stuff all day.

With the midterm elections approaching, how could the results reshape education policy and the outlook for Edtech companies and schools? 

[00:44:36] Reg Leichty: Huge, right? So I did mention earlier, right, these online safety and privacy conversation's really bipartisan. It's very bicameral. But we are facing an election on November 3rd that could really shake the snow globe, right?

We have 36 governors races, many of which are open seats, right? Where there's not an incumbent, where no matter which party is elected, there's gonna be a new face. Those governors are gonna have appointments into state boards of education and other agencies across state government. They're gonna have their own policy ideas.

They will have likely been running right now in twenty twenty-six on some of these issues of technology, including AI use, including AI data centers, which has become, as you know, right, a flashpoint, and I think a very clear picture of the broader debate about these technologies that's taking place over, you know, whether or not these data centers should be created.

So 36 governors races, House and Senate races, right? Where we might see a change of control in the House, which is very narrowly divided. We may also see a change of control in the Senate. I don't think that's gonna directionally impact Congress's work on these issues, but it's gonna be bringing new people to the table, new committee chairs who might have different perspectives.

And then we're gonna have, again, turning back to the state level, thousands, literally thousands of state legislative elections happening as well. And again, even in those very hyperlocal races, they're probably gonna be a lot of discussions about AI data centers, technology use in schools, privacy and safety of children's data and experiences.

So I think we can't underestimate how important the midterm elections are gonna be for the next three to five years of technology policy at least. 

[00:46:39] Ben Kornell: Wonderful. Well, Reg, if people wanna find out more about your services at Foresight Law and Policy, what's the best way for them to find out more? 

[00:46:46] Reg Leichty: Yeah, visit our website, flpadvisors.com.

We have a link there that you can join our weekly education policy newsletter. Check us out on LinkedIn, both the Foresight LinkedIn and my personal one. We try to write a lot about these issues. I've recently written about the expectations in the lame duck for work in this area, so would love to continue the conversation with your listeners.

[00:47:08] Ben Kornell: Thanks so much, Reg, and thank you, Edtech Insider listeners for joining in. We're gonna have more about politics and education this fall, coming up with a webinar in September, and we've got some analysis coming out in our own newsletter too. So stay tuned. Thank you all for listening, and thanks, Reg, for joining.

[00:47:25] Reg Leichty: Thanks, Ben. 

[00:47:26] Ben Kornell: All right, Edtech Insider listeners, we have a great session for you right now with CodePath, Mike Wales, and Madison McCormick. Mike is CodePath's Senior Director of AI programs, leading initiatives like the Claude Core program with Anthropic. He's a self-taught developer and Air Force veteran who went on to build top-secret applications for the NSA and Udacity's first nano degree program.

Welcome, Mike. And Madison is CodePath's strategic initiatives lead, building partnerships across industry, philanthropy, government, and higher ed to open tech pathways for first-generation low-income students. She previously led initiatives at Schmidt Futures, Global Citizen, and Comic Relief US, I'm excited for some jokes, Madison.

And holds an MA in International Relations from NYU. Welcome to the podcast, Mike and Madison. 

[00:48:19] Mike Wales: Thank you for having us. 

[00:48:21] Madison McCormick: Thank you for having us. 

[00:48:22] Ben Kornell: All right. So Madison, CodePath students are significantly more likely to land tech jobs and earn higher starting salaries, particularly those from low-income and underserved backgrounds.

What do you think CodePath is doing differently that traditional higher education has struggled to replicate at scale? 

[00:48:39] Madison McCormick: Yeah. Thanks, Ben, and great to be here. So since two thousand seventeen, CodePath has served over fifty thousand computer science students in the United States. We've primarily focused on students who come from under-resourced communities and institutions, so community colleges, state schools, the, the workforce development pipelines that serve many communities in our country.

And we're really proud to share that our alumni have graduated from our programs and earn on average, uh, twenty thousand more than their peers. They're earning as much as in their first jobs out of graduation, those at MIT or, or Stanford, our data shows. So this data reflects what we've built and what we've perfected in our classroom, which is a unique approach that many higher ed institutions have benefited from partnering with us and for credit pathways as well.

But what we do, which is a little bit different flavor than your traditional path, is we bring instructors with real recent industry. Some are actively working at these major tech companies, and they go to campus, they go into our classrooms virtually as well, and they work with our students directly. So that kind of network connectivity, but that applied learning as well, is connected early on into the learning journey.

We also do some non-technical things like pod structures and community and tech fellows and alumni connections that really build a strong relationship to learning for our students. Navigating the landscape that we are in right now in the labor market, it's really important for the relationship to learning to be sticky for our students and, and so that's something that has persisted for most of our students as they navigate through our career.

And then we're also, like I said, really focused on that applied learning. So as you advance through our curriculum, you are working on open source code for real companies that we match you with, like GitLab, and you have GitHub repositories that you graduate with. So you're, you're actually graduating with validated skills and a portfolio as you enter your career.

So I'll pause there, but that's kind of the, the overall model and where we like to meet students, especially those in low-income institutions. It's 

[00:50:40] Ben Kornell: amazing. It's like performance assessment, things we've been talking about in K-12 education for years, but actually enacted and enabled. One of the most interesting parts about your program is the Claude Core, which is an ambitious effort to place one thousand AI-trained fellows into nonprofits across the country.

And here at Edtech Insiders, we've highlighted the challenge of the nonprofit and governmental space of actually leverage-- fully leveraging AI. And we have asked the big tech companies to commit at least two percent of compute to nonprofit purposes, just so that there's the actual c-compute capability and capacity.

But also what you're doing is really it's about the skills and around training people. Mike, can you tell us a little bit about that program? What have you learned so far? And from your perspective, what skills or mindset make someone successful in using AI in the context of social impact? 

[00:51:39] Mike Wales: Yeah. Thank you.

It's a, a great question. Codepath has a long history of working with highly technical talent like, you know, computer science or engineers. And as you mentioned previously, I, I'm self-taught. Thirty-two years ago, I, I taught myself how to do this. I wish a path like Codepath existed. I had to create my own. So love being involved in this opportunity.

One of the most surprising things that we've noticed with the-- this role that fellows are going into, yes, it, it's technical, but it's more about change management and process management and understanding how people work in these technical organizations. There's a, a surprising amount of the curriculum that we've put together that is focused on the problem discovery, navigating your organization, critical thinking skills, first principles thinking.

And so, uh, these are all things that any organization would look for as, as hiring engineers, and it's just interesting to see how much of that translates to this, this AI native role that comes forward Some other surprising things that we've learned a-as we continue to build out this program is just the demand.

We knew we would have a, a lot of demand for this program, but I don't think we anticipated this level of demand, and we've received it from across the entire globe. And as we continue to assess it and address our scalability on, on our end, even at this early stage, working with our partners, then figuring out how to structure the, the program for future years, we're, we're really, really excited to see how this continues to grow.

And then I guess the, the final thing I'll mention in terms of like, you know, what has really surprised us is the demographics of the students that have come out through that assessment process, both degree holders and non-degree holders. It, it's been really exciting to, to see that, that data. This isn't, uh, you know, our application and assessment process didn't try to create this data, but coming out the other side, the successful applicants are roughly 50/50 on those that have a formal degree and those that do not.

And so it's just exciting to see that we are reaching out to that source of talent that CodePath was created to address, the underrepresented, the underserved, and making sure we hit home with them. 

[00:54:13] Ben Kornell: Amazing. What's interesting is people have had ideas like this in the past, but they've tried to go it alone.

Instead, you've really built partnerships across philanthropy, government, higher education, and industry. And Madison, this is a big part of what you do, and that often takes too much time, and it's too hard to kind of get all those pieces in motion. And yet it feels like you all have caught lightning in a bottle here, where almost the alignment is supernatural and the demand is off the charts.

Can you tell us a little bit about how all of that came together? 

[00:54:48] Madison McCormick: Yeah, lightning in a bottle is a great way to put it. The timing and the synergy between us, uh, CodePath and Anthropic was really, I think, something that made this possible, especially in an accelerated way. I've built fellowship programs with much more resources in a time when AI tools were not available, and it, it took a, a year, for example, for one fellowship.

We actually were able to build this concept based off of a trusting partnership that we already had established with Anthropic earlier in the year and bring it from concept to execution and launch within six weeks. So the timeline for even developing this program in the background is, I think, a playbook in itself of how two organizations with different capacities and resources came together.

But what we're also designing for and why it's met this, I think, this demand and the signal for what we're designing for is, is needed right now is last year was a learning year for both Anthropic and CodePath. We were seeing AI's impact on the labor market across sectors. Our early career technical students that were graduating with incredible skills were navigating a, a rapidly shifting labor market.

And Stanford recently did a study where I believe sixteen percent of early career roles have contracted. But as they're re-emerging, the expectations for these roles are a bit more full stop. They're more interdisciplinary, and they're more senior. And so our students have-- we've seen this from just how we've even adapted our own programs for our new applied AI pathways.

So all this was happening in the background. And then at the same time, on the other side of the clockwork design is the nonprofit and public sector, which overnight last year really changed the ability to not only survive but scale, where federal funding might not be available as it was before, where those nonprofits that are normally not equipped with the technical staff that were able to adapt and, and move forward quickly were falling behind very quickly.

And so we witnessed this as a nonprofit ourselves from our own operations and how we've been able to integrate AI for many years, what that's done for our quality of our programming, for our ability to scale, for our ability to really survive and thrive in this time and where our peers have, have struggled.

And so this demand on the early career side and on the public and social impact sector is something that we're designing for. And partnering with Anthropic has made this possible to bring it to scale in such a quick time. 

[00:56:55] Ben Kornell: Yeah, super impressive. I think for both of you, you're looking at today a thousand AI trained fellows for CloudCore, but the goal growing to over a hundred thousand fellows per year.

Do you believe that this narrow focus on AI experts is likely the long-term path, or will there also be ways in which the average nonprofit executive also kind of dips their toes and gets a one oh one of AI? I'm really thinking about diffusion of innovation here and how we think about unlocking AI's potential in the social impact sector.

What's your thought? 

[00:57:35] Madison McCormick: What I've witnessed, and for me as someone with more interdisciplinary background, worked in many different social impact settings and institutions, what's exciting for me right now at the moment we're in is that the technical skills are no longer sequestered only to the full-time technical staff.

It is an expectation and I think a necessity for, yes, the CEO, not just the CPO or the CTO or the COO, to understand the capacity of these tools because in order to be, I think, a visionary for your organization and, and what you can even deliver and the s- the scale or the, the depth that you are working towards, having an understanding of the tools, it's really a systems thinking architecture mindset that engineers are so well-equipped with that now can belong to everyone.

So that's my perspective, but Mike, would love to hear yours. 

[00:58:21] Mike Wales: I agree with everything you said, Madison, and I, I think where we are right now is everyone is starting to become acquainted with these tools, and now we're starting to lean towards what does AI fluency mean. 

[00:58:33] Ben Kornell: Mm-hmm. And 

[00:58:33] Mike Wales: so we're building frameworks.

We're working with our partners at Anthropic to develop these sorts of frameworks. So organizations can actually say at the individual level, but also at the organizational level of, "Okay, now that we're using these tools, here is our return on investment." And we're really excited to continue that research and figure out, like, what does AI fluency actually mean, and how does an organization tell their story around that?

[00:59:01] Ben Kornell: And to both of your points, like who is the builder has fundamentally changed. And it used to be you had to hope and pray that some developer or tech person could understand your model. Now, you know, most of our social impact entrepreneurs are naturally builders anyways, but now that, as you said, Madison, the technology building capability is no longer a barrier to entry.

So- 

[00:59:27] Madison McCormick: Mm-hmm ... 

[00:59:27] Ben Kornell: that sense of competency and growth is really important. 

[00:59:32] Madison McCormick: Yeah, and we have our playbook, of course, and, and our perspective. But what's the-- I think one of the coolest parts about Clocktower is we've developed this platform called Common. So e- when you meet the host organization and the fellow from the application stage all the way through running through the program, all of the curriculum, which will be adapted for each side week over week, it'll be personalized.

We'll be collecting a lot of data, and so it'll move from just understanding weekly average usage and more adoption numbers that people are familiar with now with AI usage to actually what Mike was saying, those validated skills within and the layers of AI fluency directly correlating with the outcomes of the impact on these institutions.

And we will have been able to done that assessment and that research across four hundred plus organizations, which will provide a, a real roadmap, a real skills wallet to validate the fellows at the end of the program, and also a playbook for nonprofits and many other sectors of what it really looks like to be AI fluent and what that does for your operations, for your budgets, for your impact 

[01:00:33] Mike Wales: One of the other things that we're really interested in exploring is just-in-time curriculum.

So within the Commons platform that Madison mentioned, we are going to be gathering, you know, the host organization's product roadmaps, feedback from the fellows, from the mentors themselves. So the curriculum that a fellow is going through week by week is going to be influenced and unique to that particular fellow, their organization, the challenges they're going through, as recently as the challenges that they discovered last Friday, for example.

So every week, fellows are getting something uniquely tailored to them and the-- their organization and the challenges they're focused on, and we're really excited to see how this just-in-time curriculum attributes to success to both not only the, the fellows themselves, but the host organizations and their outcomes.

[01:01:32] Ben Kornell: Yeah. It's fabulous, and it's so meta that you're training people to use AI, and you yourself are using AI to be more dynamic and responsive. I love that. So we've done a little bit of past and present. Let's do future. Five years from now, what do you imagine this whole ecosystem looks like? And what will, you know, your role be in enabling that future?

Of course, we're techno-optimists here at Edtech Insiders, but I'm curious what your vision would be five years from now of this intersection between AI and social impact in the nonprofit space. 

[01:02:08] Mike Wales: I'll go first, and I'll, I'll kick it over to Madison. I think for us at CodePath, it's proving that this apprenticeship and internship sort of model works and has an impact on the, the lab-labor market that we anticipate.

There's two sides to that labor market. You have, you know, supply and demand. So the fellows, we wanna make sure that the AI fluency that I, that I mentioned, does that actually contribute to additive earnings for those fellows? Do they continue to earn that twenty K more that, that Madison mentioned earlier, or even better?

And then also, uh, increased agency in their own careers. Does this make them more able to move around the career market, you know, of their own accord? On the other side, the demand side on-- from our host organizations, we wanna see, you know, does the fellow's involvement in their organization, does that change the way that they work?

Does it improve their hiring, their budgeting? Do they operate differently after that fellowship ends? And so we're really excited to continue this research and, and see how things shake out on both sides of the labor market, and I think it's one of the key reasons why Codepath and Anthropic pairing together and partnering together here, a great partnership that has been really influential in both of these-- both sides of the market.

[01:03:30] Madison McCormick: Yeah. And then I would say five years or even earlier, we're extremely ambitious on both sides. So we've committed to building this to be borrowed. So everything that underpins what we're doing is collecting a robust amount of data that we can learn from, and that we would like to provide as a playbook for larger adoption.

So if this gets federal level adoption for programs like an AmeriCorps, that would be something we're interested in supporting. We've also looked already at other sector partners like state governments and small businesses and even enterprise, what it looks like to design for those industries. The demand that Mike mentioned at the beginning of the call, we've received tens and tens and tens of thousands of applications just for the first cohort of one hundred.

So there is more than we and two partners can really serve in a year, and so we are open sourcing as much as we can as soon as we validate it that it is the right model and that the curriculum is achieving those outcomes. The goal is for much larger adoption and maybe not even within the US. 

[01:04:30] Ben Kornell: Yeah, I love it.

And, and you know, what we're seeing in the Edtech space is the emergence of micro companies. This idea that you can grow and scale your impact as an Edtech organization with ten people to ten million ARR. That's a business framing on it, but I think for social impact organizations, many of our social impact organizations are small but mighty teams.

And if you think about the ability for them to amplify and extend their work, either by having all of the back office and, like, routine elements that take up a bunch of their time taken care of by agentic AI or by having extensions in their service and their reach and their access through productized methodologies, I think it's a really exciting time to be in this social impact space and be a social impact entrepreneur.

And then I love your vision around the fellows. You know, my biggest hope is that those fellows, some of them get reabsorbed back into tech and AI and can actually bring the depth and meaning of their experience with social impact to the models themselves so that we realize that optimizing ad throughput or creating addictive, engaging AI companions isn't the only future for this technology, and it actually has some transformational impact benefit.

[01:05:52] Madison McCormick: Yeah, Ben, you addressed a lot of the things we're really hopeful and curious for, especially on both sides. But if nonprofits end up changing their budget or their structure, the way that they have even formatted their teams to maintain that quality of scale, go deeper, go wider, that's something that we'll be observing.

So thank you for mentioning that. We're on the same page. 

[01:06:11] Ben Kornell: Well, Mike and Madison, it's been such a pleasure to learn more about CodePath. If others who listen to this want to learn about CodePath and follow up, what's the best way for them to find out more? 

[01:06:22] Madison McCormick: A great way to get in touch with CodePath is to contact us via our website, codepath.org.

You can find a contact form. 

[01:06:28] Ben Kornell: Awesome. Thanks so much for joining us, Mike and Madison from codepath.org. And we're gonna be back with more on your journey as you start breaking barriers in nonprofit AI. Thanks so much for joining us today. Thank you. Very 

[01:06:42] Madison McCormick: exciting. Thanks, 

[01:06:43] Alex Sarlin: Ben. Thanks for listening to this episode of Edtech Insiders.

If you like the podcast, remember to rate it and share it with others in the Edtech community. For those who want even more Edtech Insider, subscribe to the free Edtech Insiders newsletter on Substack.

This season of Edtech Insiders is brought to you by Starbridge. Every year, K-12 districts and higher ed institutions spend over half a trillion dollars, but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, and then helps you turn them into personalized outreach fast.

Win the deal before it hits the RFP stage. That's how top Edtech teams stay ahead.