← Edtech Insiders
Edtech Insiders
The Future of Tutoring Is Human + AI: Sam Olivieri & Daniel Halper of Step Up Tutoring
00:25:51
Loading summary
Transcript33 lines
- [00:00]
A
This season of EdTech Insiders is brought to you by Cooley LLP. Cooley is the go to law firm for education and ed tech 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. This season of EdTech Insiders is brought to you by Starbridge. Every year, K12 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.
- [00:53]
B
So there's 600,000 students annually who are being paid through a work study job. Most of these are on campus jobs working in food service. You're at the campus bookstore. They're pretty menial labor. They have limited value in terms of relevant skill building or career pathways. But there are certain aspects of that federal work study program that can be leveraged for improving the quality of work study jobs.
- [01:25]
A
Welcome to EdTech Insiders, the top podcast covering the education technology industry. From funding rounds to impact to AI developments across early childhood, K12, higher ed and work, you'll find it all here at EdTech Insiders.
- [01:41]
C
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 backchannel insights from Alex and Ben. Hope you enjoy today's pod. Hello EdTech Insider listeners. I am here once again with an incredible team of entrepreneurs, Sam Olivieri and Daniel Halper from Step Up Tutoring. Sam is the CEO and she began her career as a special education teacher and has spent over 20 years expanding educational opportunity including as Chief Strategy Officer at Great Schools, Partner and Entangled Solutions, Friend of the POD and Senior Principal at Guild Education. She holds a BA from Pomona and an MPP from ucla. Go Bruins. No. Let's go Cardinal. Daniel Halper is a co Founder of Step Up Tutoring and lead step up AI Labs, the organization's R&D arm building AI enabled tools to strengthen tutor practice. A Stanford product design grad, Good job Daniel. Go card. He was previously a founding AI UX Engineer at Chiron Learning and has taught with KIPP Wise Readers to leaders and Stanford's preschool counts. Without further ado, welcome to EdTech Insiders.
- [03:09]
B
Sam and Daniel, Thank You, Ben. Thanks for having us.
- [03:12]
C
It's so great to have you both on. For those of you following along, I've mentioned Step Up Tutoring. I don't know, it must have been like two years ago. And we continue to follow this incredible journey of this nonprofit that really came out of a national tutoring movement. So let's start at the beginning. Sam, you've spent over two decades expanding educational opportunities. What convinced you that this particular model at Step up is the right one to scale nationally at a time where many tutoring efforts struggle to expand?
- [03:44]
B
Yeah, absolutely. Well, I mean, first, just starting with the concept of tutoring. Tutoring in and of itself is not a novel idea, right? Tutoring has been something we've been doing for millennia. And the evidence base on tutoring is probably as close to ironclad as you can get. Looking across ed tech interventions and education, whether you're looking at the Bloom's two sigma study or the decades of research coming out around high impact tutoring showing that this adds consistently anywhere from 3 to 15 months of additional learning. Like this is an intervention that is tried and true when we know that it works. Pairing a child with a one on one trained tutor or small group tutoring is one of the best investments in terms of return on investment that we can get for our dollars in education. The problem that I've always seen with tutoring is that for hundreds of years this has really been used as a tool to help families with means help their kids get ahead. And Step up is really the idea is how to flip that on its head. How do you use this? How do you create a model for tutoring that can really scale effectively and effectively. So this can be something that closes gaps in access to education, not expands them. And so we have had this opportunity in the tutoring space with the expansion, especially in the post Covid world around tutoring being offered through school districts, high dosage or high impact tutoring being delivered through districts, a lot of Esser money obviously as we all know spent in this space and still challenge with this is that it's expensive. Quality high impact tutoring interventions still run over $2,000 per student annually. And so you still have this situation where still roughly like 1 in 10 kids who need it are actually able to get this. So the way that we've been looking at it is how do we actually leverage assets within our community to create a solution for tutoring that can scale? Because what there is not a shortage of is caring adults. There are so many opportunities for folks who want to make a difference. The problem is cost and complexity. Like how do you actually connect these caring adults to the kids who need it? And we found kind of where are these pockets of people? So whether you have volunteers that are looking for opportunities to give back, college students who are looking to build their experience profile, or aspiring teachers who want to build their experience, this is actually a massive underutilized workforce of mission driven and fully subsidized labor. So we've been working on how do you actually capture this Building a national tutoring corps of volunteers of work, study paid college students and aspiring teachers and connect them through a virtual one on one delivery and then pair them with the right technology, including our AI tool suite that we've been working on to actually address their experience gap and build their instructional skillset to help a novice tutor actually deliver high impact tutoring with efficacy. But it's all built around this core idea that relationships are what actually make tutoring most effective and relationship scale through people. So we just need the tools to actually make that work.
- [06:47]
C
Well, that resonates so much. And I think we've kind of come back to tried and true known elements of education and tutoring where that relational connectivity is just as important as all of the other levers, Instructional design, delivery, dosage, et cetera. Daniel, when you started this, you were actually a senior at Stanford. I'm curious what made you kind of see the need and the possibility of this national tutoring core, but also why did you decide to make this a nonprofit at the time where we saw huge rounds for other companies that have since grown, blown up and then faded? How did this start? And why go with a nonprofit model?
- [07:29]
D
Yeah, it's a good question. And as you said, I was a senior in college studying product design in Silicon Valley. So as you can imagine, everybody in that major at Stanford was trying to start something. I mean, everyone was trying to build the next great. I really thought a nonprofit made the most sense given the model we were trying to implement. But yeah, it started really as like a grassroots movement to just support students in need. We started up very quick. Covid hit, schools closed, just the world was on fire. And I mean, it sucks as a college student to not be able to work in person and have that college experience. But for these young students, it hit much harder. I mean, they were not getting any personalized attention at home. They weren't getting. There's a massive amount of learning loss. I'm sure most of your viewers already know about this, but there's just so many reports coming in about just how bad the state of education was, especially in those early days of the pandemic. So I had a lot of experience kind of working in low income communities, doing tutoring in classrooms and I was sitting at home with nothing to do with like most other people at that time. So it was pretty natural for me to want to get involved and be part of this Solution. And to LUSD's credit, they had rolled out devices, like 200,000 plus devices to students. They rolled out broadband Internet, they put the infrastructure in these homes so that a program like ours could even come about. And on the other end of the coin you had millions of adults stuck at home looking for something to do. They wanted support. They just. Everyone there's just the news was so grim at the time. And yeah, we launched Step Up Tutoring. I mean it started as just one on one tutoring and we were hoping to just get people there to show up to some sessions twice a week and hopefully get like some level of momentum. But people started and they didn't stop. I mean we got great traction. We did almost no marketing. We had over like 500 signups in weeks and they completely ignored the three month commitment. People stuck with it for the full year. They had formed these relationships. So it started very much as kind of a call to action during a very tough time, but it obviously grew into something much more so that kind
- [09:42]
C
of teacher core or national tutoring corps, that idea really caught fire at the time. But there's this big challenge of like how do you take a volunteer who maybe has had no experience and set them up for success over time. And, and this is actually where part of why we wanted to have you on the show where we're really curious about how you're using AI. We see a lot of the for profit tutoring companies going to AI as the tutor. This is another example of where you're going a different way. Can the two of you talk a little bit about the role of AI for the tutor, not the student?
- [10:17]
D
Yeah, to be transparent, I'm very bullish on AI and I think AI as its own has enormous capabilities in the education space. But I think having a tutor layered on top of that adds something that AI alone often misses, which is accountability, engagement, motivation. Right now only 5% of edtech licenses are actually used. So you have these great products that honestly look like magic. We would never believed our eyes, we saw these in even 2022, but you just don't have them get being used. But tutors I think offer something unique they're human. It doesn't always feel like this way. But students respect humans a lot more than they respect AI and they're going to show up, they're going to be more engaged. A lot of our tutors are also students from similar backgrounds as the students that they're tutoring. So they're able to relate in a way. And it's just a whole different paradigm. When you have like a human show up every week, it creates that accountability, that connection. There's some sort of like social contract there. We've seen really good attendance. I mean, our students show up to their sessions and they show up for not just like a few weeks, they show up for the school year.
- [11:28]
C
I want to get back to the AI piece, but Sam, you both have made a compelling argument that the demand is incredibly high. The need is super high, but the supply is the challenge. And I can understand during COVID everybody sitting at home, that's one kind of supply of tutors. But today you've got work study students and retirees who have busy other things going on. What are the biggest barriers for you to scale that supply side? Are they policy barriers? Are they operational barriers? If we wanted everybody in America to tutor two hours per week, what could get us there?
- [12:05]
B
Well, work study has been a really unique opportunity here. I'll say we started, you know, it's Daniel and the founding team really started as a volunteer corps. But when we realize that. So there's 600,000 students annually who are being paid through a work study job. Most of these are on campus jobs working in food service. You're at the campus bookstore. They're pretty menial labor. They have limited value in terms of relevant skill building or career pathways. But there are certain aspects of that federal work study program that can be leveraged for improving the quality of work study jobs. One point is seven and a half percent of work study dollars have to be allocated to community service roles. That's a requirement of the funding. So the colleges have to find community service placements for at least seven and a half percent of those dollars. Some colleges have actually taken the voluntary step of expanding that to 15%. So they say we are going to meet a call to action to have 15% of work study dollars go to community service roles. Tutoring and tutoring within Title 1 schools is the biggest single placement for those community service roles. The first thing we could do would be increase those percentages. Allocate a larger percentage of these are existing federal funds. They go to help kids pay for college, but they could be better used to make the jobs that they're doing add more value to us from a societal perspective. Secondly, we could look at how to make those jobs more meaningful to the students so actually have embedding or incentivizing or requiring certain percentage of those dollars to go to true work based learning opportunities that are also career aligned and thinking about pathways to teaching as one of those opportunities. So are you actually having career aligned relevant skill building, helping these students build durable skills with these jobs that don't just help them pay for college but actually help them build a skill set and experience profile that will help them get their next job. And so those are, there are proof points for this and these could be built upon. But I think there is a meaningful opportunity to. You could think about if we could tap 5% of college students today to do this kind of work that could add a meaningful footprint to supporting low income elementary school students across the country.
- [14:32]
C
Well and if you think about the employment levels coming out of college, having more relevant skills or even like a teacher pipeline, there's a secondary effect of more jobs or more job opportunities for students. It's really, really fascinating. So then that really puts the spotlight around effective preparation and support for your teachers. If the two of you could get into the nitty gritty, what does AI support for a tutor look like pre, during and post session? Let's use your work study example. It's a college student who's on work study. What would they be expecting? How would they be using AI?
- [15:13]
A
We'll be right back. 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@higheredu.com that's h I r e e d u.com
- [15:52]
D
yeah. So before a student even technically joins the session, we have a little session briefing. They see it's an analysis of their previous session with that student. And we had noticed just on average students were showing up early to these sessions or tutors were showing up early, students were showing up a few minutes late. Maybe we have a quick recap of what happened the last session. We did a whole research project with Stanford around talk moves. So we have these a set of talk moves that are really highly effective in driving up tutoring quality and we recommend specific talk moves based on what was used or what was actually not used in the previous session. We're able to analyze the video for missed opportunities to use these talk moves. And they can go into and double click and see exactly what point in the transcript did this show up where. Where they could have used it, but also just kind of get a quick skim if they don't want to dive too deeply into it. They can also then see what they covered on the curriculum level, what topics they covered, what the student was struggling with. Our tutors are often covering three or four students at once, so this gives them a much better overview and just reminder of where that student's at. And then in the live session, once they have actually joined with the student, they will kind of have this tool there at hand at any time. It'll generate problem sets for them specifically stranded around these priority skills. So the highest leverage skills that student needs to know for that grade level. And we customize it for their specific curriculum. So if the student is in Eureka, we're going to be kind of generating problems that align to that. If they're in Illustrative Math, we'll generate problems that are kind of aligned to Illustrative. And then we actually give these tutors, again, because they're novice tutors, for the most part, we give them a really good breakdown of how to explain this problem, how to guide a student through it, then also how to support a student who might already have some foundational knowledge.
- [17:48]
C
And so you've like integrated or ingested all of those curricula into your system or how does that input work?
- [17:55]
D
Yeah, we're in the process of fully building out some of the diversifying the curriculum. But Eureka Math, we've been able to basically pull in additional context.
- [18:05]
C
Wow.
- [18:06]
D
For every priority skill. So we've gotten it really, really solidly one to one with kind of how you would see a Eureka lesson in their own materials. Illustrative, similar thing. We're also building out a knowledge graph or we're leveraging actually learning commons knowledge graph to really try to ground our curriculum in a much larger data set outside of just the priority skills that we're pushing. Yeah. Down the line we will do maybe some fine tuning, custom, like building custom models. But for now, just pulling in the right context for that specific lesson has been really effective. These are open source curriculums, so the AI already has pretty good foundational understanding of kind of how Illustrative or Eureka works. We're actually analyzing their whole video before the session, so we're able to see exactly what the student and tutor are saying, but also what they're drawing on the board and how they're saying things, what the student looks like. If the student seems confused, we're able to capture all that nuance and that has made the analysis a lot better. So talk, move analysis. But also when deciding what curriculum and problems to present to the student in their next session, we're able to really understand and track that student's mastery of these topics.
- [19:17]
C
That's amazing. Like so many people talk about AI being an assistant teacher and they have a model of the AI teaches and helps with practice. But in this model, it's like the AI is a coach for the tutor to help the tutor be more effective.
- [19:34]
D
Yeah.
- [19:34]
C
One thing that I've just blown away, you have a small, nimble, nonprofit team and yet you've been able to do all of this five, 10 years ago wouldn't have even been possible technologically. Sam, I'm just curious, given your footprint today, given what you're doing with AI, what does this mean? If you're able to hit your goals over the next five or 10 years, what does that vision look like? And maybe just on the number side, how many tutoring sessions have you delivered so far? Where do you think it could go?
- [20:04]
B
One thing I want to call out about what Daniel just emphasized about what Daniel just said is that this is something we've approached this from a sort of a learning and developmental standpoint. So we set out first to really focus on tutor feedback. We thought, what's a great application for AI? And we said, hey, we could give actually really robust personalized feedback to tutors and really help them deploy affect the principles of high impact tutoring. Well. And then as we began to do that, we realized, hey, there's actually a bigger opportunity here. Not just to help the tutors deliver their tutoring better, but actually to help deliver better content help not just how they're delivering, but what they are delivering. Because we realize there's an opportunity to solve this long standing problem in tutoring, which is curricular alignment. And we know there's so much evidence around the importance of coherence in what kids are learning, but you don't know how. Eureka math teaches multiplication of fractions. That's really hard to do well. But when you train our AI on the pedagogy embedded in Eureka, actually it can solve for that. Then we realize, hey, there's actually a bigger opportunity here, which is actually to use the AI to assess the kids mastery. We can actually do assessment in the flow of Learning, we don't have to pause learning and shift towards assessment. There's a lot of just different opportunities that we have kind of realized by just engaging in this work and how fast the models are developing, kind of keeping an open mind to what else is possible. If we are really clear about what we're trying to solve for what are the problems that our tutors are facing in making sure that we're solving for real problems that come out of our user base and we're keeping close tabs on sort of what we see as the right functionality while then working on things like reducing the processing costs so we can actually do this at scale. So the opportunity here is to actually have this as something that we can do at scale. And if so, this really could be a cost effective way to deliver high quality tutoring that could be on par with that Bloom's two sigma study. If you really can build a pair, this novice, caring adult who really is coming at this with a purpose, this isn't gig work. These are young people who are doing this for a true purpose and they could be doing a lot of things with their life, but they've decided this is what they want to do. The number one thing that motivates our tutors is knowing that they're having an impact. So pairing them with the tools that help them feel and be effective, that's kind of the sweet spot of actually helping them be effective. So right now we've delivered over 250,000 hours of one on one tutoring. This year we're serving, we have about 3,000 tutors in our core right now. They're active right now, served about 5,000 students this year and we've doubled year over year for the last two years. So we are really on track to get this to be something that really can serve 50,000 plus students a year. And if we can continue this to piece together both are honing the quality and making sure that this is something that is delivering the results with efficacy while also working on kind of the systems and policy change to make this kind of opportunity to be something that is the default, not the exception, then we could really have the vision of a national tutoring course where there's an effective, trained, caring one on one tutor that's available for every kid need.
- [23:28]
C
Gosh, I love that vision and also just it's making me think about what AI is unlocking in a new perspective. Just because we so often think about for profit models, we think about direct student engagement, but this idea of people who have a high intent around making an impact, getting the right scaffolding and support to deliver on that from wherever they live. That is just really inspiring. I think our listeners are going to want to learn more about Step Up Tutoring as an organization and a model and maybe they want to volunteer too. Where should people go to find out more about Step Up?
- [24:07]
B
Come to our website, stepuptutoring.org it's a super meaningful opportunity. I tutor, Daniel tutors. A lot of us tutor. My kids are paired with tutors. This is something that you don't need a lot of experience and you don't need a lot of time, but you can make a huge difference and also have a lot of fun working directly with the kid, really getting to know them and learning together. It really is super fun. This is something we think there could be a tutor for every child. Also, anybody can tutor. As a parent, I'll say it's the age old question of I'm trying to help my kids with their homework and they're like, mom, that's not how the teachers do it. That's not how they teach us. And this tool actually helps us to get that right, helps us solve for that. So we could use these tools to make them available to parents, to caregivers, anybody who is directly working with a child and help them be effective in their support.
- [25:00]
C
Well, it sounds so appealing. I'm going to check it out. I know my mom, Janet Lowery, when you see her come across Ben's mom signing up to tutor. Thanks so much. Sam Olivieri and Daniel Halper. Step Up Tutoring. It's one of those organizations that I think continues to inspire us with what's possible. Thank you so much for joining EdTech Insiders.
- [25:23]
B
Thanks. We got your mom's name down, so we'll look out for her.
- [25:26]
C
Sounds great. She's our original listener to the pod, so I'm sure she's listening right now.
- [25:32]
B
Early adopter. That's what we love. Love an early adopter.
- [25:35]
C
All right, thanks so much for joining us today. Bye bye, guys.
- [25:37]
B
Thanks so much, Ben.
- [25:39]
A
Thanks for listening to this episode of EdTech Insight. 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. 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 to. 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 edtech company or buying one, contact Tuck Advisors now.