Throwdown on AI in Education: Insights and Q&A

hi everyone I know people are trickling in into the zoom um and maybe we&#39;ll give everyone 10 to 20 seconds to to settle in but today we&#39;re really gathered here today for a very very speci special one-of aind epic Throwdown where we discuss the good the bad the ugly in Ai and education um this webinar is hosted by our startup Academy which is a series that gsv puts together to support edtech entrepreneurs in our ecosystem you can check out all of our previous content in our YouTube channel as well as any other content from our recent asgc Summit and the air show on our YouTube as well a massive massive thank you to are sponsors cooly for making this webinar and this series possible I know a ton of the startups in the audience are very appreciative for the structure today we&#39;ll do about 30 minutes of conversation I don&#39;t know how we&#39;ll be able to distill so many ideas and Concepts and debate into that 30 minutes um but we&#39;ll also have 15 minutes of audience Q&amp;A at the end so please please if you&#39;re in the audience add any questions you have for Dan for me for Ben in the chat and we&#39;ll get to them in the last 15 minutes I want to introduce our two amazing speakers today who are in our brawl our bull today is Ben Cornell he is the managing director of the common sense of Common Sense growth a social impact investment fund that amplifies the mission of Common Sense Media Ben is also the co-founder of edtech insiders which I&#39;m sure many of you tune into it&#39;s the largest edtech podcast Pleasant newsletter plus community and you can check it all out on all podcast channels as well as edtech Insiders .org our bear today is Dan Meyer he is the director of research at amplify he previously taught high school math to students who didn&#39;t like high school math and after earning his Doctorate from Stanford in math education he joined Desmos to design Desmos math he&#39;s also the author of the very awesome newsletter math worlds which you can find as Dan Meer sub Dan meyer. substack in the space and the conversation around Ai and ettech is much much more Nuance than just those two stances but for the sake of a very fun brawl and debate that we&#39;re about to have I&#39;d love for both of you to take the position of a bull and a bear and share high level opening statements on why each of you are optimistic or skeptical about AI in attech as the conversation progresses I know um the the conversation will evolve and you do not by any means have to feel obligated to stay strictly within your role of bull or bear but I will kick off with the bull then hi everyone it&#39;s so great to be here Ben Cornell uh I started my teacher my experience as a a Teach for America teacher taught for five years loved every moment of it and really my personal mission has been how do we scale what works in education I think every uh everybody in their life has had a great teacher how might we give access to all learners um to great teaching and learning how might we improve the efficiency and delivery of it and then how might we improve outcomes so that every learner can reach his or her potential I&#39;m super excited to be here with Dan uh Avid Reader of of your newsletter and also just been really following your train of thinking around quality and what teaching and learning looks like and I think overall our our space does need to ship the conversation from the tech to the Ed it&#39;s called Ed tech for a reason Ed comes first tech comes second and it reminds me of the time when we had iPads and it&#39;s like everybody needs an iPad what for not really sure but we need one and so I&#39;m very excited at this moment about the Practical use cases in education um big picture I will just say we&#39;re in a third Revolution from agrarian to Industrial um there was a lot of change in education the big change in agrarian to Industrial was this kind of localized some get it others do not totally bespoke to a standardized quality of Education that is accessible to all and we should applaud those at the turn of the century who said education is a right that every child should have unfortunately um you know our world has evolved so rapidly that a floor based model where you say there&#39;s the same kind of standard it doesn&#39;t really allow kids to thrive in a dynamic world and also that factory model system really had Sorting Hat elements where it would categorize students into below basic you know uh meeting and exceeding in some way shape or form that was really modeled to go with worker mid-level management and Senior leadership and that is just not a world that is Equitable it&#39;s not a world that reflects uh the way career go and it doesn&#39;t reflect our values as a society so what is this new era and why is it exciting the new era which really is around skill and competencies kind of the evolution of humans is learning how to learn not learning specific facts to do a job this aih AG is really about creating adaptive Learners where there is highly distributed highly personalized adaptive learning opportunities that are in the zone of proximal development for every learner at every phase of their journey and also from an efficiency standpoint we really have an opportunity to optimize our scarcest resource which is Educator capacity so how might we take um the challenge of a teacher shortage crisis and turn that into an opportunity to really leverage the humans to do what they most uniquely do and allow technology to automate tasks and that automation of you know tasks is actually what technology has been doing for a long long time even when you go back to The Agrarian era so that is a through foot where I&#39;m most excited about technology is five areas and then there&#39;s a couple that I&#39;m studying one is assessment ultimately if we change assessment we change what the system does No Child Left Behind taught us that actually systems of assessment can dramatically change how teaching and learning works and now we have the opportunity to have adaptive real-time assessment that is instrumented through technology tools powered by AI in a way that is supportive of Educators and Learners number two is Educator efficiency that can be anything from lesson planning and grading but it&#39;s this amount of the teacher work time that can be condensed to enable them to do the more human tasks when I was teaching I remember I would assign a 10-page essay I have 150 students that&#39;s 1,500 pages that I&#39;m going to have to grade and my turnaround time would be so long that the uh feedback would be irrelevant the third thing we&#39;re excited about is learner practice so right now ai is not very good at teaching novel Concepts introducing a new um learning standard or or creating the initial why behind the learning but in terms of going to the learning gym and getting the Reps doing math practice doing reading practice Writing Practice there is a lot of extensibility of the tech to help kids build those muscles and so we&#39;re seeing a bunch of examples we&#39;ll talk about today fourth is around data in infrastructure and insights and um you know what we have in education systems is a lot of data but unfortunately only 5% of the data is structured data that&#39;s easy to assess those are test scores and so on we have all of this other data that is you know a student poster on the wall or a dialogue between two students that is very hard for today&#39;s instruments to capture but there&#39;s an opportunity with AI which is really good at capturing unstructured data and turning it into structured data and then separately we have all these Data Systems that don&#39;t talk to each other it&#39;s apples and oranges and now we have a technology that&#39;s really good at um organizing data so that it&#39;s interoperable so now we can have oranges and oranges talking to each other and creating Data Insights and then the fifth is parent to school engagement I&#39;d actually say this is the area where I&#39;ve been most under impressed but um in terms of what we learned in covid is that parents and educators are a great team in supporting Learners and there&#39;s an opportunity for us to build Bridges whether that&#39;s across language divides whether that&#39;s information flow whether it&#39;s advice I think there&#39;s an opportunity for T parents to be a greater part of our system what you&#39;ll find about those five areas is that these are not new areas that we&#39;ve been struggling with just in the last year these are age-old challenges assessment teacher efficiency student practice data data interoperability infrastructure and parent engagement it&#39;s just finally we&#39;re at the point where the technology is starting to intersect with those and I think there&#39;s three or four other areas things like mental health students with learning differences where I&#39;m still in the holding Camp because they&#39;re so high stakes in those areas that having a I as a part of those systems may work but I worry about like where where we draw the line between the human oversight and the the AI oversight but also great promising technology that we can talk about today I&#39;ll flip it over to you Dan right on let&#39;s go uh thanks Ben appreciate your comprehensive work with edtech insiders as well keeping us all updated and informed on things um in the space I&#39;m coming to you folks as a lover of mathematics uh and it&#39;s depth and creativity uh of teaching and that unique relationship around oriented around learning and a student&#39;s personality as well and also of students how what an amazing thing it is a huge responsibility to have all this time and influence on kids um over an entire year and if I come across as prickly uh in this event or on my substack it is not out of an abundance of anger or anything like that it&#39;s more just an abundance of love for uh this work and everything inside of it I&#39;d say I&#39;m coming here um agreeing with Ben in lots of ways this is a technology that will get lots of use over the next few years and onwards um there&#39;s going to be some successful businesses built on top of it uh you know teachers will eventually use it every day in the same if only in the background of other products for sure I&#39;m coming here with a lot of skepticism towards anyone who believes that this will be a revolution in learning lots of folks use that word I&#39;m defining Revolution here as anything that would change um the trajectory of learner attainment uh the order of say a covid like a seveno drop is what we saw in e8th grade math scores like we&#39;re not going to see that based on this Tech or anything close to it or a revolution in um in teacher retention for instance is often like we&#39;ll give them gen AI chatbot uh teacher assistance we are also not going to see this is not the answer to our teacher shortage um concerns and perhaps Ben and I can unpack that later my reasons for saying this is that um geni is one of a series of Technologies all oriented around personalized learning quote unquote um technologies that include as far back as you know movies film reels TV the radio certainly YouTube uh was promised to be like you can you can watch your own YouTube video you want to watch you can get autog graded exercises which will tell you the next video to watch all of these Technologies including gen AI which is for sure the best most polished uh most interesting new rendering of those Technologies they&#39;ve all been built on the same faulty ideas about what education is and requires and I I perhaps some of you share these ideas you might disagree with me but some of those faulty ideas include things like schooling is about cognition primarily that&#39;s false schooling is as much about socialization and learning to be a human being um other ideas include for instance that um students are interested in spending extended periods of time with them and their laptops doing different things from their peers time and again uh we saw this throughout covid um time and again we see that students are much more interested in doing things socially another faulty assumption is that students see their classmates as liabilities like isn&#39;t it awful you&#39;re in a classroom with 30 other kids who must have completely distinct nonoverlapping needs from the teacher I&#39;m and again we see students say no that is not the problem for me that you edtech operators think it is I do not want to be separated from my friends our differences are navigable we can work this out and it&#39;s more fun when we do so I see gen as being won in a long lineage of personalized learning Technologies again definitely the best of the bunch so far um but uh I just say like if you were to go back in time to last spring and tell yourself what the usage looks like among students and teachers of gen AI right now I think you&#39;d surprise yourself back then like back then you know people were saying well this is new as of this school year came out in like mid semester break or whatever and surely next year will be the year that this thing really kicks kicks off kicks kicks into gear and we&#39;re seeing here now there there might be some like major product Market fit issue with Gen given the usage numbers we see so I&#39;m excited to explore that and and also the question of whether we should be doing personalized learning at all with all of you folks thanks everybody yeah so much to discuss and I&#39;d actually really love to touch on that exact kind of last Point um around just AI hype versus adoption so common sense media actually just very recently put out an awesome survey on just how students are using gen um and interestingly over half 51% of young people age 14 to 22 have tried gen but only 4% use it daily um they&#39;re excited about gen but worry about job loss and IP theft Etc um how do you both interpret this disparity or in usage and and awareness and adoption I&#39;ll it off to you Ben first yeah um well first I don&#39;t think um I I would love to dialogue a little bit about personalized learning in part because I think there&#39;s a confusion between individualized and personalized and modes of instruction for personalized learning include whole class small group individual near peer and the best models that pedagogically Foster learning use all of those modes of instructions and therefore the best Ed tech tech following Ed again really enables those um situations and supports those situations um and there&#39;s some good examples out there on the on the utilization of generative AI I&#39;m a elected School Board member in St Carlos and by day I talk about Ai and all of its possibilities and and at night I say whoa let&#39;s slow it down and I think there&#39;s a couple reasons one this technology is built for adults this is not an under 13 use case period full stop and in the under 18 it is really recommended for adult supervision and so ultimately in our schools with 18 years and under students we should be really thinking first about adult utilization of the technology and then prescribed use cases of AI for a pedagogical purpose and not for AI for AI sake second I think that um some of the research just showed how unaware some students are that they may actually be using AI or adaptive products without even knowing them and I think that&#39;s problematic I think that being aware of what you&#39;re interacting with and how maybe an algorithm is potentially stereotyping or creating bias or making guesses that&#39;s something that user should have a right to know whether they&#39;re adults or whether they&#39;re children but especially with children who are who may not be positioned to accurately evaluate Ani&#39;s response and so um you know I found the results actually kind of encouraging from a common sense standpoint and from an AI standpoint which is we&#39;ve got this new technology it&#39;s accelerating so fast but we have this adaptive change which is how do we incorporate it into teaching and learning and ultimately the systems that transform pair technical change and adaptive change together you can look at Healthcare as an example um where every new innovation in healthcare comes with research and protocols to roll that out and then what you see is decade on decade hospitals are better such that the best hospital 50 years ago wouldn&#39;t be allowed to operate today best school 50 years ago in many ways operates the same so why what is the metric that we should be looking at I think it&#39;s more about practitioner use Educators and administrators and how they&#39;re using it in K12 that&#39;s the leading signal of whether it&#39;s adding value and on the learner side I&#39;m hopeful that the AI is underneath the hood but is more visible around where the AI is so that we uh we can teach them to be um comprehensive and thoughtful users of AI in their daily lives yeah I don&#39;t have a whole lot to argue with here I don&#39;t think with Ben I would just I I ask myself like I want to question anytime I hear the premise that like how should we integrate AI into institutions I I don&#39;t take it as self-evident as most I think that AI should be integrated not for ethical or moral concerns though those are certainly real just in terms of like is this actually useful like teachers are extremely resourceful and if and if you you were able to sincerely and legitimately promise them like a 4 Hour you know or 40% or some kind of meaningful difference in their prep time or their ease of facilitating a diverse classroom of students for instance they&#39;re not going to turn that down like that&#39;s Ben and I have both been teachers and know like it&#39;s really any port in a storm at a certain point like whatever you can help me with um so I I tend to like ease off considerably any sense of obligation towards teachers like or institutions to say like how how can you do this thing as though it is self-evident it should be done and instead to to Wonder first again Ed before the tech what are the the real impressing needs of teachers and students and school leaders and can geni play a role there versus going in a priori and saying like gen plays a role here how can we get you to see that yeah I almost feel like that ties into you had a piece a couple weeks ago talking about how a lot of edtech platforms just you know they they their positive results are based off about 5% of the students um and a lot of times they they blame students for not engaging enough with these adpe products and there&#39;s almost kind of this uh mirroring view that we have to put the the burden on teachers to upscale themselves on and you must use this AI tool so um curious how you all are thinking about kind of that institutional change management um but also thinking about the ways that AI might continue to exacerbate this kind of skew and existing inequities in edtech platform usage yeah so the question here is about how how edtech might skew existing uh inequities like that&#39;s a that&#39;s a a great question and yeah what I think these uh different research these efficacy studies summarized by Lawrence Hol found is that oftentimes the positive results which they all will show um emerge from the 5% of users who use the product the most um and they these companies for instance KH Academy or zern or others would toss out the bottom 95% in terms of usage so in the KH Academy case for instance 5% of Learners in this study where teachers opted their students in um use KH Academy at the rate of 30 plus minutes per week and um you know we there&#39;s we can only investigate the counterfactual so much we can only know so much about these different students but I find it very easy personally to believe as uh that those 5% of students are the students who least need academic support and I would just name myself and I won&#39;t speak for Ben here but I was a s student who would take whatever resource was provided to me and use it at whatever level was asked for and probably benefit from it but I would also have learned you know just fine from a ham sandwich so what we see I think time and again is that EDC operators are have a deep infatuation with their product as they must in order to survive in difficult uh times economically um but that lends themselves to a lack of criticality a lack of self-reflection on why their their tool their product fails to meet the needs of what is potentially um the neediest 9 5% of Learners and the worst part to me is just seeing how some of these Founders when they when they&#39;re are are asked about that that disparity then turn the responsibility for meeting the needs of the 95% back onto the teachers or essentialize the students in ways that are quite demeaning like well these kids just don&#39;t want to engage um that kind of spirit and kids and teachers just really deserve a whole lot more than that yeah and if I might flip that question um especially for for you Ben especially as someone who invests in a lot of ettech startups um because we are in a startup Academy we do have a lot of Founders and audience what strategies might you suggest for AI at Tech or just generally edtech startups to you know effectively collaborate with teachers in school so that this 5% and 95% Dynamic doesn&#39;t happen yeah first let&#39;s just talk about the 95 versus 5 I think you know Google is the largest edtech company in the world let&#39;s it&#39;s a rounding error for Google but let&#39;s be real edtech is more expansive than the narrow set of things that Lawrence who did a great investigation was looking at and I think in particular what Lawrence published is relevant to the tutoring industry and this idea that an AI tutor is going to solve all problems and I think we who&#39;ve been in the space understand that tutoring does not occur with an open blank where you say I want to learn this today and it teaches you something that&#39;s literally not how tutoring Works tutoring works with a an educator who understands what a learner needs to work on collaborates with that learner and comes up with a plan and the opportunity for AI is extending the surface area for practice but it is not the to displace the educator or the tutor in that um coming back to like how educator um um edtech companies should think cover your ears Claire but one thing is they should ignore their investors and invest in professional development and implementation support and by the way there are plenty of studies that show professional development is a waste of money and time too but what what I&#39;ve seen and what I think there are good case studies is when you pair your technology with good implementation and support and alignment vertical alignment with the administrator the the teacher and the learner you can have meaningful outcomes and so in my expansive definition of edtech I would also include curriculum companies like amplify for whom as as Dan you commented before it&#39;s table Stakes to make sure things are implemented well those are actually really good examples of things that have shifted access to high quality content have enabled better long-term engagement and ultimately better outcomes and that&#39;s got to be we&#39;ve got to double down on that now why is it a good business decision well I think one thing to realize is 80% of our spend in districts is on human beings about 10% is on facilities so we&#39;re really just talking about this last 10% and a meaningful part between four and 8% is spent on professional development and curriculum and 1% is spent on te Tech and so if you&#39;re trying to sell your edte tool as a software tool you&#39;re just really going in at that 1% level but if you&#39;re able to be connected with core learning or you&#39;re being able to connect with professional development there&#39;s actually real Revenue there that behaves like recurring Revenue because we as school leaders know you know we have to invest in rolling out our products the last thing I will say is the current market is a recession Market we are all in my school district districts across it is not more is more less is more right now we want coherence and these Bottoms Up sales strategies where you&#39;re doing um premium product adoption you&#39;re getting the 5% of students finding value you&#39;re also getting 5% of teachers who are finding the value you&#39;re not finding ways to be valuable to all Educators in often in those Bottoms Up motions um and obviously there&#39;s some good exceptions where there&#39;s schoolwide adoption but you need to have a vertical alignment too where you&#39;re working with the school leaders to say what are the five priorities of the school there&#39;s only so much schools can take on in addition to serving everybody lunches in addition to providing transportation in addition to being a community center so um I just would say invest in Services it&#39;s time yeah I just i&#39; love love add on to that there from the operator side like I helped uh build up Desmos classroom acquired by amplify as a product that has extremely high NPS High retention scores High retention rates and I would just say that key to that success a lot of different things many authors but we hired very well um we had um people who are product people Specialists CTO caos we had Educators and designers and Eng all Under One Roof all with um a lot of respect for one another&#39;s disciplines um but also a lot of good instincts about them as well so we hired Engineers who were former teachers teachers who had CS backgrounds at times um that was uh extremely key and I see right now especially with AI edtech startups where the tech demands are so high higher than say like the just kind of the full stack apps of your um that a lot of the founders have just extremely technical backgrounds a lot of passion for what they think education is um but are just sorely in need U of some ballast on the teaching experience side as an assessment I would just say ask yourself as a founder or a leader like do you know the answer or have your answer to these questions like in in muscle reflex memory what makes teaching effective what makes Learners feel good in their education what makes teachers feel powerful if you can&#39;t answer those uh there&#39;s there&#39;s more learning that needs to happen uh for you and your team that is both you know both of you have highlighted very important points um you know as a venture investor I&#39;m General bias towards answering a question what does the world look like if all this does go right and we&#39;ve talked a lot about um where this could go wrong and there definitely more questions than than answers right now but um similar to a lot of the Q&amp;A from the audience it looks like they&#39;d love to hear what do you both see as opportunities in the space and what&#39;s actually getting you excited and looking forward to the future of AI and at Tech I I&#39;ll I&#39;ll have one first that&#39;s all right Ben I got a relatively short answer and I you&#39;ve seen so much of the space love to hear what you have to say about it but um I I I have just like a very I might be wrong but a very clear well-developed for me answer to what is the work of teaching which is to invite from students what they have brought with them to school and then help them develop it and the deal is is that in any class of 30 there are just some really interesting and Rich and uh differently formed ideas about what&#39;s going on and when the teacher successfully invites that from students students feel warm and at home it feels hospitable they are engaged um but then the teacher has this job job of taking all these interesting ideas ideas which aren&#39;t neatly captured by a frequency distribution graph of multiple choice responses like they&#39;re they&#39;re like they&#39;re very they&#39;re Loosely formed um and then to make meaning out of that and so when you when you think about like how do I take 30 thoughts and find their commonalities think about what question would be best next think about who are different exemplars of these different clusters of ideas thinking about which students haven&#39;t been called out recently for having a a really interesting idea who might benefit uh from feeling called for doing something right um I start to see a lot of possibilities for Advanced Technologies like geni or other kinds of AI to help teachers make sense of what is going on in the average classroom yeah just just to build on that um I would I would put a word on what Dan was talking about which is assessment and really finding a way to enable better and more Dynamic assessment and I talked a little bit about this earlier first from a impact standpoint we measure what matters most to us and when No Child Left Behind was launched in my classroom and in my school we went from English history science math Arts band electives to three periods of English and two periods of math my instruction was you have these far below basic kids and Below basic kids get them up one more rung and even more fine tuned than that I was given a spreadsheet of those who were considered cusp teach to those Learners so as much as I think we can throw shots at Ai and what it can or can&#39;t do continuing with our status quo system is also untenable and we need to be investing and adopting new strategies and that&#39;s why I believe assessment is at the root of all of this if we change what we measure we will not only create direct change but indirect change where Educators can do a lot of the strategies Dan was talking about to show Competency Based learning to do more Dynamic um work in a learner zone of proximal development or their groups or their classes zone of proximal development and if you look at Healthcare and that history that I talked about with adaptive and Technical change what preceded that it was the diagnostic reev Revolution it used to be actually a total art a dark art uh Healthcare was but once there was this ability to look under a microscope and analyze someone&#39;s blood and then send those labs and get it back in 24 hours you have a no do reflect Dynamic that allows for constant diagnosis and constant testing and obviously you know Healthcare has not been equitably distributed at all and there&#39;s still a lot of room to go but that ability to have real-time adaptive formative assessment versus lagging thepoke annual assessment is the fundamental Dynamic that will shift our systems now last on the business side people pay for assessment look at irating look at uh etss look at SATs look at any of the test prep companies where the money is in our system is in assessment and even the curriculum companies understand that their curriculum is only sold in the context of the assessments that it exists in so wherever you are in the in the system whatever Learners you&#39;re serving make sure you&#39;re anchoring your work on a dynamic adaptive assessment and I think that&#39;s how you move the ball forward collectively and individually feel like that&#39;s a amazing point to to wrap up the kind of structured Q&amp;A part of our conversation I know there are tons of questions flowing in so would definitely love to get to those um but very quickly would love to turn it to Naomi from coie our awesome sponsors for this webinar uh for for quick chat hello everyone and thank you Claire and Dan and Ben uh my name is Naomi May I&#39;m an attorney who has the pleasure of working with hundreds of Education sector clients here at kie LLP for those of you who don&#39;t know us yet we are a global law firm with the most robust fullservice edtech practice in the world um our te IT Tech Team loves to facilitate conversations like those we&#39;re engaging in today um and not surprisingly they mirror many of the conversations that are happening right now on the legal and the legislative fronts um policy makers also executive branch Regulators are similarly wrestling through these issues um on the one hand they don&#39;t want to miss something bad but they also don&#39;t want to stifle that very real promise um this dissonance though honestly is exactly what I think we should embrace because I think it&#39;s where the real learning where the best collaboration can occur and where we can really find the best Solutions together as a community so before we head back to Q&amp;A um I wanted to give you a quick overview of what&#39;s happening with AI regulation generally um the US Congress has four to five task forces across houses and executive agencies are also engaged but the real action right now was really at the state level where we have seen a recent slew of activity early State AI laws um have the potential I think to exert a real outsize influence in the trajectory of of AI regulation in the US um the focus is heavily heavily on consumer protection U Utah was the first US state to enact um a broad consumer protection statute specifically governing AI Colorado followed by enacting its own comprehensive AI regulation the Colorado AI act that was signed into law last month it&#39;s going into effect next year and for upcoming proposals we have our eyes on Connecticut&#39;s bill as well as California both seem to be gaining traction um but despite all the uncertainty in the in this space um there are some commonalities to all of these regulatory schemes which I think are worth keeping in mind as you&#39;re you&#39;re building your companies and your products the primary one is notice notice notice um of how you&#39;re using user information and specific and clear disclosures on the front end for example if you&#39;re training on consumer data your users need to know um other Topline recommendations would include um determining the applicability of state laws and again not just um AI laws but also the dozens of data privacy laws particularly children&#39;s and student data privacy laws and increasingly laws you might not think would necessarily apply to AI State wiretapping laws is in area we&#39;re seeing a lot of activity right now um also monitoring regulator engagement and enforcement activities understand how those factor into your risk profile and then also um regularly monitor AI outputs are they false are they misleading are they discriminatory you need to know so that&#39;s just tip at Iceberg but I know you want to get back to the main show so I will send it back to Claire and our panelists thank you so much Naomi super helpful and I&#39;m sure many in the audience will continue to have questions about navigating this very complex and involving legal landscape um so onto the Q&amp;A we have a ton of questions and I might do a little bit of merging um one question we have is I I&#39;ll merge these two the first one being at what age do you feel is appropriate for students to start interacting with geni and two another uh audience member mentioned that their daughter uses Desmos her algebra class uses the CPM curriculum uh they appreciate the use of former teachers in developing the application and really see good uses for that and um think it would be really beneficial to get student feedback along the same lines Ben mentioned gen being for adults not for younger than 18 or 14 should miners take an active role in shaping gen for their age group uh I pass to B on that one yeah yeah I I&#39;ll be I&#39;ll be brief and also big fans of Desmos and Eli and and just you know the the engagement of that product um when I say that AI is for adults you know first I think it it also the Nuance here is that products that use AI are already in the hands of kids as as young as you know two three four years old and at Common Sense we really recommend that parents push off screen time in general like digital products as long as they can but now that we have omn modal you also have like voice activated products um I think there&#39;s a difference between direct engagement with a generative AI open model like a chat TBT versus indirect when it&#39;s incorporated into a product in age appropriate ways so one of the companies I&#39;ve invested in is Ella which is a digital reading assistant it&#39;s a side-by-side reader that enables that reading practice it&#39;s not to replace a reading teacher but it&#39;s really to make sure that students have an interactive 20 to 30 minutes of reading each day and that they&#39;ve got an AI coach helping them they&#39;ve done the work to tune and train that model um to be appropriate for that task and it uses speech and so um you know I&#39;d say when I talk to parents when I talk to Educators I think one we have to acknowledge the reality that kids are going to be engaging in digital tools that have ai but two that we as adults can do a lot to understand the intentionality in which it&#39;s used and the and the Frameworks that that both unlock uh features and capabilities and those that restrict awesome um for Dan a bunch of people have highlighted how much they appreciate um what you mentioned around hiring well um and a couple of Founders have mentioned either themselves being from Tech backgrounds or or education backgrounds asking for forums or spaces what do you suggest or or places where um these Founders can find Dev or product folks to find educational advisers or people in the classroom to find technical collaborators or or thought Partners I think the the best thing that these people can do is to seek to volunteer and provide immediate value in classrooms near them odds are very good that if you&#39;re a technical founder or a Founder in general or just a college graduate that your own experience in schools mirrors um the experience of your target user very little like they&#39;re just they could not be more dissimilar so getting in into classrooms that speak more towards the 95% of of students and teachers out there would be my first recommendation and what we found is that when teachers feel like you understand their needs when you feel like they feel like you&#39;re you&#39;re building tools not just to make life easier but to make them more powerful to help them do better the work that they want to be doing um but they they tend up block and gravitate towards you and and offer themselves willingly as collaborators um the first the first job is to understand what what their work is um in order to support them better yeah um and on the note of you know engaging with schools uh some of the audience members are from K12 environment uh they&#39;re mentioning how schools are struggling a ton with funding partly obviously due to eer funds expiring and enrollment generally going down with the Advent of AI do you all think that schools and districts should receive funding targeted for responsible use for AI through training and products um and how can District leaders best engage with the tech innovators in the space yeah I I&#39;ll jump in on that one as a school board member and so on um I think it would be a mistake to provide funding for AI I think would be great to provide funding for new modes of assessment or new modes of learner practice or new modes of um engaging parents um I also think that the way when I talk about Technical and adaptive change the way we socialize Innovation and education is not everybody adopts overnight or it&#39;s not we throw a huge contract together we do pilots and we are Learners as Educators and then we understand how in our context we would Implement that what I why I&#39;m so optimistic about AI is in part because the abil ability to deliver some solutions uh we we can instrument things in ways that we never have before but also we can operate our organizations or deliver those tools at a much lower cost so one of the cost barriers that you know all schools are dealing with is Sr funds are gone enrollment is down and our cost basis is higher so if you can provide the same or better value for a tenth of the cost schools and districts are interested um and then on the on the um kind of role of the Federal Government and or the role of states or funders I think we need much more workaround Frameworks for evaluating tools and um this is probably an area where Dan is a leading advocate for uh efficacy and evidence to support procurement right now I&#39;m I literally know everything about head Tech like I&#39;m in I&#39;m in edtech insiders and we have to we have votes to adopt uh products and I don&#39;t know whether it&#39;s effective in our schools or for our Learners and I&#39;m like experiencing like a a sales process where I should have all this inside information and is still incredibly opaque so guidelines around what to use what not to use what questions to ask um what does quality evidence look like what is it not so um you know throwing money at this problem RIT large or throwing money at AI specifically mistake throwing it at the the areas we want to improve and then figuring out how AI might intersect that is a good idea yeah I hard agree uh just that we are way too early to earmark funds for this particular solution like it needs to be proven out through efficacy study funded by self-funded by the corporations or funded by the government at NSF or philanthropic foundations would be fine but we&#39;re just we funding problem spaces uh sounds like a great move here and specific Solutions is uh premature I&#39;d be pretty pretty shocked to see that push through yeah um and then I think we have a question from one of the audience members that mentioned that as Ben says we measure what matters most to us um and one thing that matters to a lot of communities and Families is kids social and emotional wellness and skills the social learning that Dan you reference um and that currently it feels like there aren&#39;t really great measures uh of those skills not just environment and it seems like there aren&#39;t Great Tech solutions for this um so the question is is that due to neglect or is that just fundamentally impossible for Tech and can AI come in here well I think that the um there are some decent measures here where ran did a study commissioned by The Gates Foundation about personalized learning implementations Nationwide huge study and they had a really solid set of measures looking at student Social Development asking them like do you know an adult here do you feel like a valued member of your community how safe do you feel um items measuring those constructs I&#39;ll go back and look at and look at those and especially look at the finding that in the personalized learning schools where to Ben&#39;s point it involves more than Tech but certainly heavy usage of technology in this way that students um had diminished feeling of uh Social Development of belonging of safety in those environments and there are examples of technology um that support the social unit like our technology is not designed for sustained periods um of onetoone device usage when students are on those devices um what they are thinking about goes to the teacher ideally to support a conversation in the whole class where students see one another&#39;s value in that conversation there&#39;s other kinds of models that support that so just like an agree that that there both very important and also just to say hey some stuff is already happening has happened and is happening awesome and then I know we are slightly over time but just the last one and we&#39;ve seen quite a few of these uh for both of you what are AI education use cases that have you excited or ones that you love and they they might just be features or or entire products um very quickly just one to two as we wrap up um maybe Ben first sure well I mentioned l already um I think magic school is doing a lot to um understand AI can do so many things for you in a school you actually have a scatter plot of use cases and needs among Educators and so it&#39;s this ability to provide um opportunities for every educator to find their entry point by leveraging AI for real problems that they have and a deal is also uh an educator and a school leader founder um one that I really excited about is called early bird learning you might not have heard of them they do dyslexia assessment for three and four year olds and what we had to do before is wait till children learn to or did not learn to read to assess them for dyslexia we actually have the opportunity to eradicate dyslexia as a long-term issue for Learners because when treated at age 345 you can actually um overcome all the symptoms of Dyslexia and by by working with students with those strategies so it&#39;s a game based uh you know physical location no reading required but it uses some of the same kind of tech that like a dual lingo uses to really optimize the assessment of Dyslexia and then provide supports so that&#39;s one I&#39;m super excited to have no more dyslexia in the world yeah I would just jump quickly to say that um there&#39;s a lot I like I think I find myself a little bit unexcited by um like lot lots of little chat GPT rappers Point solutions to all these different problems that don&#39;t add up to more than their sum their parts that&#39;s uh my bias there I do like teachfx I think they have a very strong idea about what teaching and learning is I admire their Clarity of vision they don&#39;t use gen but a different kind of uh supervised learning um I think it&#39;s uh interesting mostly again because they have just a very strong point of view an uncommonly strong point of view on the work of teaching and learning awesome and I know we are up that time and there are so many questions um but thank you both thank you Dan thank you Ben for your time today thank you Naomi um thank you all for the insights you&#39;ve shared today I feel like there&#39;s so much for us to Noodle on as we move forward and it&#39;ll be interesting if we can even do this you know this debate and Showdown a year later and see what&#39;s changed um and Dan&#39;s email is deer amplify.com if you want to connect with him and Ben what&#39;s your email it&#39;s Ben Cornell gmail.com and you can also check out edtech Insiders on substack check out Dan newsletter on substack and there it has all the details of how to become a subscriber and how to get you know continue this conversation awesome and Naomi if anyone has legal related questions what&#39;s your email easy andy.com amazing awesome thank you all again for this conversation and thanks the audience for tuning in thanks

hi everyone I know people are trickling in into the zoom um and maybe we&amp;#39;ll give everyone 10 to 20 seconds to to settle in but today we&amp;#39;re really gathered here today for a very very speci special one-of aind epic Throwdown where we discuss the good the bad the ugly in Ai and education um this webinar is hosted by our startup Academy which is a series that gsv puts together to support edtech entrepreneurs in our ecosystem you can check out all of our previous content in our YouTube channel as well as any other content from our recent asgc Summit and the air show on our YouTube as well a massive massive thank you to are sponsors cooly for making this webinar and this series possible I know a ton of the startups in the audience are very appreciative for the structure today we&amp;#39;ll do about 30 minutes of conversation I don&amp;#39;t know how we&amp;#39;ll be able to distill so many ideas and Concepts and debate into that 30 minutes um but we&amp;#39;ll also have 15 minutes of audience Q&amp;amp;A at the end so please please if you&amp;#39;re in the audience add any questions you have for Dan for me for Ben in the chat and we&amp;#39;ll get to them in the last 15 minutes I want to introduce our two amazing speakers today who are in our brawl our bull today is Ben Cornell he is the managing director of the common sense of Common Sense growth a social impact investment fund that amplifies the mission of Common Sense Media Ben is also the co-founder of edtech insiders which I&amp;#39;m sure many of you tune into it&amp;#39;s the largest edtech podcast Pleasant newsletter plus community and you can check it all out on all podcast channels as well as edtech Insiders .org our bear today is Dan Meyer he is the director of research at amplify he previously taught high school math to students who didn&amp;#39;t like high school math and after earning his Doctorate from Stanford in math education he joined Desmos to design Desmos math he&amp;#39;s also the author of the very awesome newsletter math worlds which you can find as Dan Meer sub Dan meyer. substack in the space and the conversation around Ai and ettech is much much more Nuance than just those two stances but for the sake of a very fun brawl and debate that we&amp;#39;re about to have I&amp;#39;d love for both of you to take the position of a bull and a bear and share high level opening statements on why each of you are optimistic or skeptical about AI in attech as the conversation progresses I know um the the conversation will evolve and you do not by any means have to feel obligated to stay strictly within your role of bull or bear but I will kick off with the bull then hi everyone it&amp;#39;s so great to be here Ben Cornell uh I started my teacher my experience as a a Teach for America teacher taught for five years loved every moment of it and really my personal mission has been how do we scale what works in education I think every uh everybody in their life has had a great teacher how might we give access to all learners um to great teaching and learning how might we improve the efficiency and delivery of it and then how might we improve outcomes so that every learner can reach his or her potential I&amp;#39;m super excited to be here with Dan uh Avid Reader of of your newsletter and also just been really following your train of thinking around quality and what teaching and learning looks like and I think overall our our space does need to ship the conversation from the tech to the Ed it&amp;#39;s called Ed tech for a reason Ed comes first tech comes second and it reminds me of the time when we had iPads and it&amp;#39;s like everybody needs an iPad what for not really sure but we need one and so I&amp;#39;m very excited at this moment about the Practical use cases in education um big picture I will just say we&amp;#39;re in a third Revolution from agrarian to Industrial um there was a lot of change in education the big change in agrarian to Industrial was this kind of localized some get it others do not totally bespoke to a standardized quality of Education that is accessible to all and we should applaud those at the turn of the century who said education is a right that every child should have unfortunately um you know our world has evolved so rapidly that a floor based model where you say there&amp;#39;s the same kind of standard it doesn&amp;#39;t really allow kids to thrive in a dynamic world and also that factory model system really had Sorting Hat elements where it would categorize students into below basic you know uh meeting and exceeding in some way shape or form that was really modeled to go with worker mid-level management and Senior leadership and that is just not a world that is Equitable it&amp;#39;s not a world that reflects uh the way career go and it doesn&amp;#39;t reflect our values as a society so what is this new era and why is it exciting the new era which really is around skill and competencies kind of the evolution of humans is learning how to learn not learning specific facts to do a job this aih AG is really about creating adaptive Learners where there is highly distributed highly personalized adaptive learning opportunities that are in the zone of proximal development for every learner at every phase of their journey and also from an efficiency standpoint we really have an opportunity to optimize our scarcest resource which is Educator capacity so how might we take um the challenge of a teacher shortage crisis and turn that into an opportunity to really leverage the humans to do what they most uniquely do and allow technology to automate tasks and that automation of you know tasks is actually what technology has been doing for a long long time even when you go back to The Agrarian era so that is a through foot where I&amp;#39;m most excited about technology is five areas and then there&amp;#39;s a couple that I&amp;#39;m studying one is assessment ultimately if we change assessment we change what the system does No Child Left Behind taught us that actually systems of assessment can dramatically change how teaching and learning works and now we have the opportunity to have adaptive real-time assessment that is instrumented through technology tools powered by AI in a way that is supportive of Educators and Learners number two is Educator efficiency that can be anything from lesson planning and grading but it&amp;#39;s this amount of the teacher work time that can be condensed to enable them to do the more human tasks when I was teaching I remember I would assign a 10-page essay I have 150 students that&amp;#39;s 1,500 pages that I&amp;#39;m going to have to grade and my turnaround time would be so long that the uh feedback would be irrelevant the third thing we&amp;#39;re excited about is learner practice so right now ai is not very good at teaching novel Concepts introducing a new um learning standard or or creating the initial why behind the learning but in terms of going to the learning gym and getting the Reps doing math practice doing reading practice Writing Practice there is a lot of extensibility of the tech to help kids build those muscles and so we&amp;#39;re seeing a bunch of examples we&amp;#39;ll talk about today fourth is around data in infrastructure and insights and um you know what we have in education systems is a lot of data but unfortunately only 5% of the data is structured data that&amp;#39;s easy to assess those are test scores and so on we have all of this other data that is you know a student poster on the wall or a dialogue between two students that is very hard for today&amp;#39;s instruments to capture but there&amp;#39;s an opportunity with AI which is really good at capturing unstructured data and turning it into structured data and then separately we have all these Data Systems that don&amp;#39;t talk to each other it&amp;#39;s apples and oranges and now we have a technology that&amp;#39;s really good at um organizing data so that it&amp;#39;s interoperable so now we can have oranges and oranges talking to each other and creating Data Insights and then the fifth is parent to school engagement I&amp;#39;d actually say this is the area where I&amp;#39;ve been most under impressed but um in terms of what we learned in covid is that parents and educators are a great team in supporting Learners and there&amp;#39;s an opportunity for us to build Bridges whether that&amp;#39;s across language divides whether that&amp;#39;s information flow whether it&amp;#39;s advice I think there&amp;#39;s an opportunity for T parents to be a greater part of our system what you&amp;#39;ll find about those five areas is that these are not new areas that we&amp;#39;ve been struggling with just in the last year these are age-old challenges assessment teacher efficiency student practice data data interoperability infrastructure and parent engagement it&amp;#39;s just finally we&amp;#39;re at the point where the technology is starting to intersect with those and I think there&amp;#39;s three or four other areas things like mental health students with learning differences where I&amp;#39;m still in the holding Camp because they&amp;#39;re so high stakes in those areas that having a I as a part of those systems may work but I worry about like where where we draw the line between the human oversight and the the AI oversight but also great promising technology that we can talk about today I&amp;#39;ll flip it over to you Dan right on let&amp;#39;s go uh thanks Ben appreciate your comprehensive work with edtech insiders as well keeping us all updated and informed on things um in the space I&amp;#39;m coming to you folks as a lover of mathematics uh and it&amp;#39;s depth and creativity uh of teaching and that unique relationship around oriented around learning and a student&amp;#39;s personality as well and also of students how what an amazing thing it is a huge responsibility to have all this time and influence on kids um over an entire year and if I come across as prickly uh in this event or on my substack it is not out of an abundance of anger or anything like that it&amp;#39;s more just an abundance of love for uh this work and everything inside of it I&amp;#39;d say I&amp;#39;m coming here um agreeing with Ben in lots of ways this is a technology that will get lots of use over the next few years and onwards um there&amp;#39;s going to be some successful businesses built on top of it uh you know teachers will eventually use it every day in the same if only in the background of other products for sure I&amp;#39;m coming here with a lot of skepticism towards anyone who believes that this will be a revolution in learning lots of folks use that word I&amp;#39;m defining Revolution here as anything that would change um the trajectory of learner attainment uh the order of say a covid like a seveno drop is what we saw in e8th grade math scores like we&amp;#39;re not going to see that based on this Tech or anything close to it or a revolution in um in teacher retention for instance is often like we&amp;#39;ll give them gen AI chatbot uh teacher assistance we are also not going to see this is not the answer to our teacher shortage um concerns and perhaps Ben and I can unpack that later my reasons for saying this is that um geni is one of a series of Technologies all oriented around personalized learning quote unquote um technologies that include as far back as you know movies film reels TV the radio certainly YouTube uh was promised to be like you can you can watch your own YouTube video you want to watch you can get autog graded exercises which will tell you the next video to watch all of these Technologies including gen AI which is for sure the best most polished uh most interesting new rendering of those Technologies they&amp;#39;ve all been built on the same faulty ideas about what education is and requires and I I perhaps some of you share these ideas you might disagree with me but some of those faulty ideas include things like schooling is about cognition primarily that&amp;#39;s false schooling is as much about socialization and learning to be a human being um other ideas include for instance that um students are interested in spending extended periods of time with them and their laptops doing different things from their peers time and again uh we saw this throughout covid um time and again we see that students are much more interested in doing things socially another faulty assumption is that students see their classmates as liabilities like isn&amp;#39;t it awful you&amp;#39;re in a classroom with 30 other kids who must have completely distinct nonoverlapping needs from the teacher I&amp;#39;m and again we see students say no that is not the problem for me that you edtech operators think it is I do not want to be separated from my friends our differences are navigable we can work this out and it&amp;#39;s more fun when we do so I see gen as being won in a long lineage of personalized learning Technologies again definitely the best of the bunch so far um but uh I just say like if you were to go back in time to last spring and tell yourself what the usage looks like among students and teachers of gen AI right now I think you&amp;#39;d surprise yourself back then like back then you know people were saying well this is new as of this school year came out in like mid semester break or whatever and surely next year will be the year that this thing really kicks kicks off kicks kicks into gear and we&amp;#39;re seeing here now there there might be some like major product Market fit issue with Gen given the usage numbers we see so I&amp;#39;m excited to explore that and and also the question of whether we should be doing personalized learning at all with all of you folks thanks everybody yeah so much to discuss and I&amp;#39;d actually really love to touch on that exact kind of last Point um around just AI hype versus adoption so common sense media actually just very recently put out an awesome survey on just how students are using gen um and interestingly over half 51% of young people age 14 to 22 have tried gen but only 4% use it daily um they&amp;#39;re excited about gen but worry about job loss and IP theft Etc um how do you both interpret this disparity or in usage and and awareness and adoption I&amp;#39;ll it off to you Ben first yeah um well first I don&amp;#39;t think um I I would love to dialogue a little bit about personalized learning in part because I think there&amp;#39;s a confusion between individualized and personalized and modes of instruction for personalized learning include whole class small group individual near peer and the best models that pedagogically Foster learning use all of those modes of instructions and therefore the best Ed tech tech following Ed again really enables those um situations and supports those situations um and there&amp;#39;s some good examples out there on the on the utilization of generative AI I&amp;#39;m a elected School Board member in St Carlos and by day I talk about Ai and all of its possibilities and and at night I say whoa let&amp;#39;s slow it down and I think there&amp;#39;s a couple reasons one this technology is built for adults this is not an under 13 use case period full stop and in the under 18 it is really recommended for adult supervision and so ultimately in our schools with 18 years and under students we should be really thinking first about adult utilization of the technology and then prescribed use cases of AI for a pedagogical purpose and not for AI for AI sake second I think that um some of the research just showed how unaware some students are that they may actually be using AI or adaptive products without even knowing them and I think that&amp;#39;s problematic I think that being aware of what you&amp;#39;re interacting with and how maybe an algorithm is potentially stereotyping or creating bias or making guesses that&amp;#39;s something that user should have a right to know whether they&amp;#39;re adults or whether they&amp;#39;re children but especially with children who are who may not be positioned to accurately evaluate Ani&amp;#39;s response and so um you know I found the results actually kind of encouraging from a common sense standpoint and from an AI standpoint which is we&amp;#39;ve got this new technology it&amp;#39;s accelerating so fast but we have this adaptive change which is how do we incorporate it into teaching and learning and ultimately the systems that transform pair technical change and adaptive change together you can look at Healthcare as an example um where every new innovation in healthcare comes with research and protocols to roll that out and then what you see is decade on decade hospitals are better such that the best hospital 50 years ago wouldn&amp;#39;t be allowed to operate today best school 50 years ago in many ways operates the same so why what is the metric that we should be looking at I think it&amp;#39;s more about practitioner use Educators and administrators and how they&amp;#39;re using it in K12 that&amp;#39;s the leading signal of whether it&amp;#39;s adding value and on the learner side I&amp;#39;m hopeful that the AI is underneath the hood but is more visible around where the AI is so that we uh we can teach them to be um comprehensive and thoughtful users of AI in their daily lives yeah I don&amp;#39;t have a whole lot to argue with here I don&amp;#39;t think with Ben I would just I I ask myself like I want to question anytime I hear the premise that like how should we integrate AI into institutions I I don&amp;#39;t take it as self-evident as most I think that AI should be integrated not for ethical or moral concerns though those are certainly real just in terms of like is this actually useful like teachers are extremely resourceful and if and if you you were able to sincerely and legitimately promise them like a 4 Hour you know or 40% or some kind of meaningful difference in their prep time or their ease of facilitating a diverse classroom of students for instance they&amp;#39;re not going to turn that down like that&amp;#39;s Ben and I have both been teachers and know like it&amp;#39;s really any port in a storm at a certain point like whatever you can help me with um so I I tend to like ease off considerably any sense of obligation towards teachers like or institutions to say like how how can you do this thing as though it is self-evident it should be done and instead to to Wonder first again Ed before the tech what are the the real impressing needs of teachers and students and school leaders and can geni play a role there versus going in a priori and saying like gen plays a role here how can we get you to see that yeah I almost feel like that ties into you had a piece a couple weeks ago talking about how a lot of edtech platforms just you know they they their positive results are based off about 5% of the students um and a lot of times they they blame students for not engaging enough with these adpe products and there&amp;#39;s almost kind of this uh mirroring view that we have to put the the burden on teachers to upscale themselves on and you must use this AI tool so um curious how you all are thinking about kind of that institutional change management um but also thinking about the ways that AI might continue to exacerbate this kind of skew and existing inequities in edtech platform usage yeah so the question here is about how how edtech might skew existing uh inequities like that&amp;#39;s a that&amp;#39;s a a great question and yeah what I think these uh different research these efficacy studies summarized by Lawrence Hol found is that oftentimes the positive results which they all will show um emerge from the 5% of users who use the product the most um and they these companies for instance KH Academy or zern or others would toss out the bottom 95% in terms of usage so in the KH Academy case for instance 5% of Learners in this study where teachers opted their students in um use KH Academy at the rate of 30 plus minutes per week and um you know we there&amp;#39;s we can only investigate the counterfactual so much we can only know so much about these different students but I find it very easy personally to believe as uh that those 5% of students are the students who least need academic support and I would just name myself and I won&amp;#39;t speak for Ben here but I was a s student who would take whatever resource was provided to me and use it at whatever level was asked for and probably benefit from it but I would also have learned you know just fine from a ham sandwich so what we see I think time and again is that EDC operators are have a deep infatuation with their product as they must in order to survive in difficult uh times economically um but that lends themselves to a lack of criticality a lack of self-reflection on why their their tool their product fails to meet the needs of what is potentially um the neediest 9 5% of Learners and the worst part to me is just seeing how some of these Founders when they when they&amp;#39;re are are asked about that that disparity then turn the responsibility for meeting the needs of the 95% back onto the teachers or essentialize the students in ways that are quite demeaning like well these kids just don&amp;#39;t want to engage um that kind of spirit and kids and teachers just really deserve a whole lot more than that yeah and if I might flip that question um especially for for you Ben especially as someone who invests in a lot of ettech startups um because we are in a startup Academy we do have a lot of Founders and audience what strategies might you suggest for AI at Tech or just generally edtech startups to you know effectively collaborate with teachers in school so that this 5% and 95% Dynamic doesn&amp;#39;t happen yeah first let&amp;#39;s just talk about the 95 versus 5 I think you know Google is the largest edtech company in the world let&amp;#39;s it&amp;#39;s a rounding error for Google but let&amp;#39;s be real edtech is more expansive than the narrow set of things that Lawrence who did a great investigation was looking at and I think in particular what Lawrence published is relevant to the tutoring industry and this idea that an AI tutor is going to solve all problems and I think we who&amp;#39;ve been in the space understand that tutoring does not occur with an open blank where you say I want to learn this today and it teaches you something that&amp;#39;s literally not how tutoring Works tutoring works with a an educator who understands what a learner needs to work on collaborates with that learner and comes up with a plan and the opportunity for AI is extending the surface area for practice but it is not the to displace the educator or the tutor in that um coming back to like how educator um um edtech companies should think cover your ears Claire but one thing is they should ignore their investors and invest in professional development and implementation support and by the way there are plenty of studies that show professional development is a waste of money and time too but what what I&amp;#39;ve seen and what I think there are good case studies is when you pair your technology with good implementation and support and alignment vertical alignment with the administrator the the teacher and the learner you can have meaningful outcomes and so in my expansive definition of edtech I would also include curriculum companies like amplify for whom as as Dan you commented before it&amp;#39;s table Stakes to make sure things are implemented well those are actually really good examples of things that have shifted access to high quality content have enabled better long-term engagement and ultimately better outcomes and that&amp;#39;s got to be we&amp;#39;ve got to double down on that now why is it a good business decision well I think one thing to realize is 80% of our spend in districts is on human beings about 10% is on facilities so we&amp;#39;re really just talking about this last 10% and a meaningful part between four and 8% is spent on professional development and curriculum and 1% is spent on te Tech and so if you&amp;#39;re trying to sell your edte tool as a software tool you&amp;#39;re just really going in at that 1% level but if you&amp;#39;re able to be connected with core learning or you&amp;#39;re being able to connect with professional development there&amp;#39;s actually real Revenue there that behaves like recurring Revenue because we as school leaders know you know we have to invest in rolling out our products the last thing I will say is the current market is a recession Market we are all in my school district districts across it is not more is more less is more right now we want coherence and these Bottoms Up sales strategies where you&amp;#39;re doing um premium product adoption you&amp;#39;re getting the 5% of students finding value you&amp;#39;re also getting 5% of teachers who are finding the value you&amp;#39;re not finding ways to be valuable to all Educators in often in those Bottoms Up motions um and obviously there&amp;#39;s some good exceptions where there&amp;#39;s schoolwide adoption but you need to have a vertical alignment too where you&amp;#39;re working with the school leaders to say what are the five priorities of the school there&amp;#39;s only so much schools can take on in addition to serving everybody lunches in addition to providing transportation in addition to being a community center so um I just would say invest in Services it&amp;#39;s time yeah I just i&amp;#39; love love add on to that there from the operator side like I helped uh build up Desmos classroom acquired by amplify as a product that has extremely high NPS High retention scores High retention rates and I would just say that key to that success a lot of different things many authors but we hired very well um we had um people who are product people Specialists CTO caos we had Educators and designers and Eng all Under One Roof all with um a lot of respect for one another&amp;#39;s disciplines um but also a lot of good instincts about them as well so we hired Engineers who were former teachers teachers who had CS backgrounds at times um that was uh extremely key and I see right now especially with AI edtech startups where the tech demands are so high higher than say like the just kind of the full stack apps of your um that a lot of the founders have just extremely technical backgrounds a lot of passion for what they think education is um but are just sorely in need U of some ballast on the teaching experience side as an assessment I would just say ask yourself as a founder or a leader like do you know the answer or have your answer to these questions like in in muscle reflex memory what makes teaching effective what makes Learners feel good in their education what makes teachers feel powerful if you can&amp;#39;t answer those uh there&amp;#39;s there&amp;#39;s more learning that needs to happen uh for you and your team that is both you know both of you have highlighted very important points um you know as a venture investor I&amp;#39;m General bias towards answering a question what does the world look like if all this does go right and we&amp;#39;ve talked a lot about um where this could go wrong and there definitely more questions than than answers right now but um similar to a lot of the Q&amp;amp;A from the audience it looks like they&amp;#39;d love to hear what do you both see as opportunities in the space and what&amp;#39;s actually getting you excited and looking forward to the future of AI and at Tech I I&amp;#39;ll I&amp;#39;ll have one first that&amp;#39;s all right Ben I got a relatively short answer and I you&amp;#39;ve seen so much of the space love to hear what you have to say about it but um I I I have just like a very I might be wrong but a very clear well-developed for me answer to what is the work of teaching which is to invite from students what they have brought with them to school and then help them develop it and the deal is is that in any class of 30 there are just some really interesting and Rich and uh differently formed ideas about what&amp;#39;s going on and when the teacher successfully invites that from students students feel warm and at home it feels hospitable they are engaged um but then the teacher has this job job of taking all these interesting ideas ideas which aren&amp;#39;t neatly captured by a frequency distribution graph of multiple choice responses like they&amp;#39;re they&amp;#39;re like they&amp;#39;re very they&amp;#39;re Loosely formed um and then to make meaning out of that and so when you when you think about like how do I take 30 thoughts and find their commonalities think about what question would be best next think about who are different exemplars of these different clusters of ideas thinking about which students haven&amp;#39;t been called out recently for having a a really interesting idea who might benefit uh from feeling called for doing something right um I start to see a lot of possibilities for Advanced Technologies like geni or other kinds of AI to help teachers make sense of what is going on in the average classroom yeah just just to build on that um I would I would put a word on what Dan was talking about which is assessment and really finding a way to enable better and more Dynamic assessment and I talked a little bit about this earlier first from a impact standpoint we measure what matters most to us and when No Child Left Behind was launched in my classroom and in my school we went from English history science math Arts band electives to three periods of English and two periods of math my instruction was you have these far below basic kids and Below basic kids get them up one more rung and even more fine tuned than that I was given a spreadsheet of those who were considered cusp teach to those Learners so as much as I think we can throw shots at Ai and what it can or can&amp;#39;t do continuing with our status quo system is also untenable and we need to be investing and adopting new strategies and that&amp;#39;s why I believe assessment is at the root of all of this if we change what we measure we will not only create direct change but indirect change where Educators can do a lot of the strategies Dan was talking about to show Competency Based learning to do more Dynamic um work in a learner zone of proximal development or their groups or their classes zone of proximal development and if you look at Healthcare and that history that I talked about with adaptive and Technical change what preceded that it was the diagnostic reev Revolution it used to be actually a total art a dark art uh Healthcare was but once there was this ability to look under a microscope and analyze someone&amp;#39;s blood and then send those labs and get it back in 24 hours you have a no do reflect Dynamic that allows for constant diagnosis and constant testing and obviously you know Healthcare has not been equitably distributed at all and there&amp;#39;s still a lot of room to go but that ability to have real-time adaptive formative assessment versus lagging thepoke annual assessment is the fundamental Dynamic that will shift our systems now last on the business side people pay for assessment look at irating look at uh etss look at SATs look at any of the test prep companies where the money is in our system is in assessment and even the curriculum companies understand that their curriculum is only sold in the context of the assessments that it exists in so wherever you are in the in the system whatever Learners you&amp;#39;re serving make sure you&amp;#39;re anchoring your work on a dynamic adaptive assessment and I think that&amp;#39;s how you move the ball forward collectively and individually feel like that&amp;#39;s a amazing point to to wrap up the kind of structured Q&amp;amp;A part of our conversation I know there are tons of questions flowing in so would definitely love to get to those um but very quickly would love to turn it to Naomi from coie our awesome sponsors for this webinar uh for for quick chat hello everyone and thank you Claire and Dan and Ben uh my name is Naomi May I&amp;#39;m an attorney who has the pleasure of working with hundreds of Education sector clients here at kie LLP for those of you who don&amp;#39;t know us yet we are a global law firm with the most robust fullservice edtech practice in the world um our te IT Tech Team loves to facilitate conversations like those we&amp;#39;re engaging in today um and not surprisingly they mirror many of the conversations that are happening right now on the legal and the legislative fronts um policy makers also executive branch Regulators are similarly wrestling through these issues um on the one hand they don&amp;#39;t want to miss something bad but they also don&amp;#39;t want to stifle that very real promise um this dissonance though honestly is exactly what I think we should embrace because I think it&amp;#39;s where the real learning where the best collaboration can occur and where we can really find the best Solutions together as a community so before we head back to Q&amp;amp;A um I wanted to give you a quick overview of what&amp;#39;s happening with AI regulation generally um the US Congress has four to five task forces across houses and executive agencies are also engaged but the real action right now was really at the state level where we have seen a recent slew of activity early State AI laws um have the potential I think to exert a real outsize influence in the trajectory of of AI regulation in the US um the focus is heavily heavily on consumer protection U Utah was the first US state to enact um a broad consumer protection statute specifically governing AI Colorado followed by enacting its own comprehensive AI regulation the Colorado AI act that was signed into law last month it&amp;#39;s going into effect next year and for upcoming proposals we have our eyes on Connecticut&amp;#39;s bill as well as California both seem to be gaining traction um but despite all the uncertainty in the in this space um there are some commonalities to all of these regulatory schemes which I think are worth keeping in mind as you&amp;#39;re you&amp;#39;re building your companies and your products the primary one is notice notice notice um of how you&amp;#39;re using user information and specific and clear disclosures on the front end for example if you&amp;#39;re training on consumer data your users need to know um other Topline recommendations would include um determining the applicability of state laws and again not just um AI laws but also the dozens of data privacy laws particularly children&amp;#39;s and student data privacy laws and increasingly laws you might not think would necessarily apply to AI State wiretapping laws is in area we&amp;#39;re seeing a lot of activity right now um also monitoring regulator engagement and enforcement activities understand how those factor into your risk profile and then also um regularly monitor AI outputs are they false are they misleading are they discriminatory you need to know so that&amp;#39;s just tip at Iceberg but I know you want to get back to the main show so I will send it back to Claire and our panelists thank you so much Naomi super helpful and I&amp;#39;m sure many in the audience will continue to have questions about navigating this very complex and involving legal landscape um so onto the Q&amp;amp;A we have a ton of questions and I might do a little bit of merging um one question we have is I I&amp;#39;ll merge these two the first one being at what age do you feel is appropriate for students to start interacting with geni and two another uh audience member mentioned that their daughter uses Desmos her algebra class uses the CPM curriculum uh they appreciate the use of former teachers in developing the application and really see good uses for that and um think it would be really beneficial to get student feedback along the same lines Ben mentioned gen being for adults not for younger than 18 or 14 should miners take an active role in shaping gen for their age group uh I pass to B on that one yeah yeah I I&amp;#39;ll be I&amp;#39;ll be brief and also big fans of Desmos and Eli and and just you know the the engagement of that product um when I say that AI is for adults you know first I think it it also the Nuance here is that products that use AI are already in the hands of kids as as young as you know two three four years old and at Common Sense we really recommend that parents push off screen time in general like digital products as long as they can but now that we have omn modal you also have like voice activated products um I think there&amp;#39;s a difference between direct engagement with a generative AI open model like a chat TBT versus indirect when it&amp;#39;s incorporated into a product in age appropriate ways so one of the companies I&amp;#39;ve invested in is Ella which is a digital reading assistant it&amp;#39;s a side-by-side reader that enables that reading practice it&amp;#39;s not to replace a reading teacher but it&amp;#39;s really to make sure that students have an interactive 20 to 30 minutes of reading each day and that they&amp;#39;ve got an AI coach helping them they&amp;#39;ve done the work to tune and train that model um to be appropriate for that task and it uses speech and so um you know I&amp;#39;d say when I talk to parents when I talk to Educators I think one we have to acknowledge the reality that kids are going to be engaging in digital tools that have ai but two that we as adults can do a lot to understand the intentionality in which it&amp;#39;s used and the and the Frameworks that that both unlock uh features and capabilities and those that restrict awesome um for Dan a bunch of people have highlighted how much they appreciate um what you mentioned around hiring well um and a couple of Founders have mentioned either themselves being from Tech backgrounds or or education backgrounds asking for forums or spaces what do you suggest or or places where um these Founders can find Dev or product folks to find educational advisers or people in the classroom to find technical collaborators or or thought Partners I think the the best thing that these people can do is to seek to volunteer and provide immediate value in classrooms near them odds are very good that if you&amp;#39;re a technical founder or a Founder in general or just a college graduate that your own experience in schools mirrors um the experience of your target user very little like they&amp;#39;re just they could not be more dissimilar so getting in into classrooms that speak more towards the 95% of of students and teachers out there would be my first recommendation and what we found is that when teachers feel like you understand their needs when you feel like they feel like you&amp;#39;re you&amp;#39;re building tools not just to make life easier but to make them more powerful to help them do better the work that they want to be doing um but they they tend up block and gravitate towards you and and offer themselves willingly as collaborators um the first the first job is to understand what what their work is um in order to support them better yeah um and on the note of you know engaging with schools uh some of the audience members are from K12 environment uh they&amp;#39;re mentioning how schools are struggling a ton with funding partly obviously due to eer funds expiring and enrollment generally going down with the Advent of AI do you all think that schools and districts should receive funding targeted for responsible use for AI through training and products um and how can District leaders best engage with the tech innovators in the space yeah I I&amp;#39;ll jump in on that one as a school board member and so on um I think it would be a mistake to provide funding for AI I think would be great to provide funding for new modes of assessment or new modes of learner practice or new modes of um engaging parents um I also think that the way when I talk about Technical and adaptive change the way we socialize Innovation and education is not everybody adopts overnight or it&amp;#39;s not we throw a huge contract together we do pilots and we are Learners as Educators and then we understand how in our context we would Implement that what I why I&amp;#39;m so optimistic about AI is in part because the abil ability to deliver some solutions uh we we can instrument things in ways that we never have before but also we can operate our organizations or deliver those tools at a much lower cost so one of the cost barriers that you know all schools are dealing with is Sr funds are gone enrollment is down and our cost basis is higher so if you can provide the same or better value for a tenth of the cost schools and districts are interested um and then on the on the um kind of role of the Federal Government and or the role of states or funders I think we need much more workaround Frameworks for evaluating tools and um this is probably an area where Dan is a leading advocate for uh efficacy and evidence to support procurement right now I&amp;#39;m I literally know everything about head Tech like I&amp;#39;m in I&amp;#39;m in edtech insiders and we have to we have votes to adopt uh products and I don&amp;#39;t know whether it&amp;#39;s effective in our schools or for our Learners and I&amp;#39;m like experiencing like a a sales process where I should have all this inside information and is still incredibly opaque so guidelines around what to use what not to use what questions to ask um what does quality evidence look like what is it not so um you know throwing money at this problem RIT large or throwing money at AI specifically mistake throwing it at the the areas we want to improve and then figuring out how AI might intersect that is a good idea yeah I hard agree uh just that we are way too early to earmark funds for this particular solution like it needs to be proven out through efficacy study funded by self-funded by the corporations or funded by the government at NSF or philanthropic foundations would be fine but we&amp;#39;re just we funding problem spaces uh sounds like a great move here and specific Solutions is uh premature I&amp;#39;d be pretty pretty shocked to see that push through yeah um and then I think we have a question from one of the audience members that mentioned that as Ben says we measure what matters most to us um and one thing that matters to a lot of communities and Families is kids social and emotional wellness and skills the social learning that Dan you reference um and that currently it feels like there aren&amp;#39;t really great measures uh of those skills not just environment and it seems like there aren&amp;#39;t Great Tech solutions for this um so the question is is that due to neglect or is that just fundamentally impossible for Tech and can AI come in here well I think that the um there are some decent measures here where ran did a study commissioned by The Gates Foundation about personalized learning implementations Nationwide huge study and they had a really solid set of measures looking at student Social Development asking them like do you know an adult here do you feel like a valued member of your community how safe do you feel um items measuring those constructs I&amp;#39;ll go back and look at and look at those and especially look at the finding that in the personalized learning schools where to Ben&amp;#39;s point it involves more than Tech but certainly heavy usage of technology in this way that students um had diminished feeling of uh Social Development of belonging of safety in those environments and there are examples of technology um that support the social unit like our technology is not designed for sustained periods um of onetoone device usage when students are on those devices um what they are thinking about goes to the teacher ideally to support a conversation in the whole class where students see one another&amp;#39;s value in that conversation there&amp;#39;s other kinds of models that support that so just like an agree that that there both very important and also just to say hey some stuff is already happening has happened and is happening awesome and then I know we are slightly over time but just the last one and we&amp;#39;ve seen quite a few of these uh for both of you what are AI education use cases that have you excited or ones that you love and they they might just be features or or entire products um very quickly just one to two as we wrap up um maybe Ben first sure well I mentioned l already um I think magic school is doing a lot to um understand AI can do so many things for you in a school you actually have a scatter plot of use cases and needs among Educators and so it&amp;#39;s this ability to provide um opportunities for every educator to find their entry point by leveraging AI for real problems that they have and a deal is also uh an educator and a school leader founder um one that I really excited about is called early bird learning you might not have heard of them they do dyslexia assessment for three and four year olds and what we had to do before is wait till children learn to or did not learn to read to assess them for dyslexia we actually have the opportunity to eradicate dyslexia as a long-term issue for Learners because when treated at age 345 you can actually um overcome all the symptoms of Dyslexia and by by working with students with those strategies so it&amp;#39;s a game based uh you know physical location no reading required but it uses some of the same kind of tech that like a dual lingo uses to really optimize the assessment of Dyslexia and then provide supports so that&amp;#39;s one I&amp;#39;m super excited to have no more dyslexia in the world yeah I would just jump quickly to say that um there&amp;#39;s a lot I like I think I find myself a little bit unexcited by um like lot lots of little chat GPT rappers Point solutions to all these different problems that don&amp;#39;t add up to more than their sum their parts that&amp;#39;s uh my bias there I do like teachfx I think they have a very strong idea about what teaching and learning is I admire their Clarity of vision they don&amp;#39;t use gen but a different kind of uh supervised learning um I think it&amp;#39;s uh interesting mostly again because they have just a very strong point of view an uncommonly strong point of view on the work of teaching and learning awesome and I know we are up that time and there are so many questions um but thank you both thank you Dan thank you Ben for your time today thank you Naomi um thank you all for the insights you&amp;#39;ve shared today I feel like there&amp;#39;s so much for us to Noodle on as we move forward and it&amp;#39;ll be interesting if we can even do this you know this debate and Showdown a year later and see what&amp;#39;s changed um and Dan&amp;#39;s email is deer amplify.com if you want to connect with him and Ben what&amp;#39;s your email it&amp;#39;s Ben Cornell gmail.com and you can also check out edtech Insiders on substack check out Dan newsletter on substack and there it has all the details of how to become a subscriber and how to get you know continue this conversation awesome and Naomi if anyone has legal related questions what&amp;#39;s your email easy andy.com amazing awesome thank you all again for this conversation and thanks the audience for tuning in thanks

Transcript for:Throwdown on AI in Education: Insights and Q&A

Transcript for:
Throwdown on AI in Education: Insights and Q&A