Edtech Insiders

Week In Edtech, 3/12: SVB Bank Run, ChatGPT4, Google Buys Photomath, and SxSWEDU

March 12, 2023 Alex Sarlin Season 5 Episode 3
Edtech Insiders
Week In Edtech, 3/12: SVB Bank Run, ChatGPT4, Google Buys Photomath, and SxSWEDU
Show Notes Transcript

In this special, UNEDITED version of Week in Edtech, Ben, Alex and special guest co-host Matt Tower discuss:

1) The sudden demise and federal takeover of Silicon Valley Bank and what it means for Tech, Banking and Edtech.



2) Discuss the rumors of ChatGPT4 launching next week with AI-created video

3) Herald the Launch of Non-Profit ai.edu with 50 partners across tech and ed

4) Unveil the (secret) purchase of top Education App Photomath by Google

5) Discuss the on-the-ground happenings at SxSW EDU

Hello, everybody, it is a special edition of the weekend ed tech. I'm Ben Cornell with my co host, Alex Sarlin. We have special guest, Matt tower from Ed Tech thoughts. And today, breaking news just an hour ago, we had news hit around Silicon Valley Bank, their takeover by the FDIC, we just heard from Janet Yellen, then on Monday, all deposits, all accounts will be paid back in full. That doesn't go for the stockholders or people who've who've loan money, but a sigh of relief across Silicon Valley across the country across edtech. We're going to break it down for you. And then we're also going to ask the question, so what? So thanks for joining us here, Matt, maybe you could just kick us off a little bit with what happened like peps up from like, Thursday, Friday to today? Sure, and, you know, I'm going to be looking for facial cues, if I if I screw anything up, please just hop right in. So, and I've been I've been working on this to try and figure out how to how to frame it succinctly. In a nutshell, Silicon Valley Bank, one of the kind of preeminent banks to venture backed startups. In, they started on the West Coast, literally, in Silicon Valley, they're a regional bank. So they're, they're, you know, like a local bank near to wherever you dear listener live. But they they made this kind of particular focus on tech. And for a long time, that worked rather well for them. You know, tech companies get a bunch of funding in the form of cash from venture capitalists, and they deposit it with Silicon Valley Bank, Silicon Valley Bank, then is able to loan out or in, in this case, they were not able to loan out and I'll get to that in a second, that money to make a return on those dollars. And like I said, for, they were incorporated in like 1983, I think, and for 40 years, they kind of hummed along, like a regional bank. And, you know, they had other types of clients do, they had coffee shops, and you know, hairdressers and all that. But again, they have this concentration on tech. And that is what kind of undid them in the end. The reason that it's tricky to concentrate on tech is tech companies don't have the same type of cash flows as a normal business. So if you are a coffee shop, you have a fairly predictable business in the grand scheme of things, that kind of law of large data or whatever, where you are going to have certain capital needs and certain certain capital inflows. And what banks often do is they lend against that. So if you need, you know, $500,000, to put in, you know, a new to put in a new branch of your coffee shop, the bank will lend you that money. And that's kind of what keeps the banking system humming along. With tech startups, they have this weird phenomenon in the business world, where they get large sums of money, and then they don't really need loans, at least for the first 510 15 years of their existence. Because they are, they're kind of using that cash to invest in research and development. And sometimes that works, you get the Facebook's of the world that grow to be, you know, multi billion even trillion dollar companies. Sometimes it doesn't. But because they don't need loans, the same way a traditional business does. Silicon Valley Bank has to find other ways to earn what's called a spread on the their deposits. So what Silicon Valley Bank did is they invested in T bills, Treasury securities from the US government, which are some of the safest things available in the entire world. And the issue here was not that the T bills were bad. It was that they were they were bought at a very low interest rate. So if we go back to 2020, and 2021, which were historically high times for fundraising, and again, that's bringing a lot of money into Silicon Valley Bank. They're really low interest rate times. So when Silicon Valley Bank was taking those large inflows, then buying government securities at a very low interest rate, they were getting very low spread. What happened this year or over the past kind of nine months is a The Fed has been raising the interest rate across our economy and the the interest rate at which you could buy government securities. So, now you have Silicon Valley Bank with a bunch of kind of low interest securities in a rapidly increasing, increasing interest rate environment. And that kind of this is where I'm trying to find the right words for it, that kind of I'm going to call it like an inflection point, ultimately cause folks, business leaders to lose faith in the bank's ability to pay their deposits if the if somebody wanted their money back. So I, as business leader, I'm looking at this and saying, Well, you know, I have a million dollars with Silicon Valley Bank, they have a bunch of low interest securities, that you know, or have a loan with your maturation date, I might want to take my money out, in case somebody else wants to park their money elsewhere. One person does this, it's not so bad thing. Three people do this, you know, the three of us here on this call, do that. It's like, Wow, that's great. 100,000 companies, which is the number of business clients that Silicon Valley Bank has, it starts to become a real problem. And also known as a bank run. And, you know, it was fueled on Twitter, and you know, VCs, reaching out to their portfolio companies actively encouraging them to switch out. So, you know, one thing about the dynamics here that I think were particularly shocking, though, is how fast it went, you know, Wednesday, you know, Silicon Valley Bank is trading at, I don't know, six, you know, the equity is like $16 billion. They had, you know, something like 165 billion and accounts, and by Friday, they're worth nothing. And they're in receivership with the federal government. It's really that liquidity crunch. That was the challenge, because the had people been willing to wait till the treasury bonds matured, they would have had the money, but it was everyone trying to get their money out all at once. So there's a lot of finger pointing here. You know, Matt, who do you? Who do you ascribe the blame to, or what's even the lessons learned? Yeah. And there's one interesting nuance that, I think is relatively unexplored, yet, but I thought was compelling. Just thinking about it from a like, user point of view, which is, you know, in the olden days, you had to literally go to your bank to get your deposits. And, you know, if the bank across the street offered you a slightly higher interest rate, there was still a fair amount of friction to transfer your funds to them, you know, go to your old bank, tell them you want all your money, have them look at you, sadly, and then walk across the street, go to a new bank and say, Hey, I'd like to open an account, they have to do the background check, you know, yadda, yadda, yadda, you know, it could take days or even weeks to get that new, better interest rate. And with SBB. or, frankly, in today's environment, it's like, we can open bank accounts in the, you know, blink of an eye, you know, I can log into my Chase credit card portal and say, Hey, I'd like your checking account. And like, they can facilitate that remarkably quickly. And, and so I think, you know, going back to that interest rate spread that we talked about, you know, if I'm looking at my SBB account, and saying, Well, you know, I'm only earning point oh, 5% with SBB, maybe, maybe I want to go to chase or, you know, some other regional bank and say, they're offering me 1%. And I think the fact that you can do that today, in, you know, minutes, really allows for that, that kind of herd mentality that you just talked about, Ben, and I think that's important. You know, obviously, we're still learning about all this, how much of a factor that is, but, but I think like, you know, again, once the kind of vortex starts spinning, it's really hard to get it to stop. I think there are some pieces of that herd mentality that we should unpack here because for people outside of Silicon Valley, you know, Ben, you're right in the middle of it. I was there, but I'm not there anymore. I think people maybe don't understand how closely knit all of these people are There's a lot of venture capital firms, many of them are literally on the same street as each other. They are next door, they all know each other, they have outsized effects, because they each have dozens and dozens and dozens of portfolio companies, depending on the size of the, of the firm. And they're all in contact with each other. So, you know, people are pointing, one of the fingers that's been pointed a lot in what I've been reading is Peter Thiel, you know, head of founder fund, a notoriously outspoken, you know, political VC, who really early on this, in this, this run, told all his companies get get out of Silicon Valley Bank, I think, you know, the way you're describing a match, it's totally accurate, as far as I know, but it sounds very casual, you're like, well, they have a higher interest rate over here. This is not casual at all, I mean, this thing now lost billions and billions in hours, it was like, and it was like, an I mean, you're right, the speed at which the money could move is definitely a contributing factor. But I really think this is a Silicon Valley specific case, it's like it's a one industry bank. For the most part, there are other coffee shops and hairdressers, as you mentioned, but almost all the money there is that also, it's it's a one industry bank, where people were putting at least a million dollars, I think that low on the business account is a million dollars. And then the FDIC only insures a quarter million dollars. So and nobody seemed to care about that, because I think in Silicon Valley world a quarter million dollars is, you know, is chump change, it's just not that much money at all. And people never think they're going to have to hit that federal backstop. But the combination of the fast moving sort of herd mentality of Silicon Valley, and the, you know, wall of the Fed I mean of the federal government and the FDIC, really, really bizarre set of circumstances. One other quick thing to add there in a lot of the coverage I've been reading is that part of the other part of the reason Silicon Valley Bank was in so much trouble is that they got a huge influx of money over the last couple of years because of the increase like once in a lifetime increase in VC funds. And then this year, it's gone through the floor, there's very little so for a bank that is entirely leveraged on on on venture capital, if they had a ton of money come in, nobody to give it to like you said Ben nobody to lend it to because that everybody they worked with already gets the venture, they don't need loans. So they put it in something safe just to have it at least hopefully make more money than the than the interest they have to pay. And then you know, and then venture falls through the floor. No More money's coming into the bank. And at our you know, relative to where it was, and then they hit this liquidity capital crunch. And I mean, I've just never seen anything like this. And I don't know, we're in a really weird moment right now, because none of us have ever seen anything like this, the federal government, as you know, just now is saying, don't panic, don't let this be a contagious thing. Don't pull your money out of every regional bank. But, and we'll see if you know, we'll see if it works to sort of limit other bank friends, we already saw a second regional bank, close as of today. One thing jumped in here too, because I don't think that people get the full picture accurate when they say this is a Silicon Valley Bank thing. Because they tend to focus on the big dollar checks and the big VCs, but really how Silicon Valley Bank made its name, you could open a business account, and you could borrow money without a personal guarantee. And for a lot of startup founders, this was actually a game changing way to bank, most of these companies were unthinkable with a traditional bank. And so actually, there's quite a number of very, very small accounts, all of which are insured, and so on. But a number of those small accounts eventually become big accounts. Because, you know, you're basically saying I need to set up a bank account, I'm going to fundraise. And the irony is like Wells Fargo won't set up your account until you have the check. And the people won't write the check until you have the bank accounts. Okay, so there's a little bit of that dynamic there. And then the second kind of piece and I don't know if this is a something to blame people on or it's a cause, but a factor here is because they were taking risk on people without you know, the collateral, they often would require that they do all their banking with the bank. And that way they would have the full picture so many people had covenants in how they you know, set up a loan or set up a line of credit. That meant that they also had to use it for their main deposits. And, and then on top of that, sometimes you might have like, let's Say you raised around, you have 4 million in the bank, some of that money, you're using Silicon Valley Bank to put in, you know, money market accounts where you might earn some interest. So it's not just all sitting in a checking account. And I think that there's like a view that oh, you know, it's these huge pots of money, why wouldn't they have diversified it? It's actually very much because they were serving this underserved or unserved group of startup founders in the early days that they'd really become a behemoth. And then I just also say, on the personal front, you know, living here, everyone is like one degree removed from somebody who's impacted. And, you know, the stories that are resonated with me over the weekend, were founders who, you know, on Thursday, frantically, were trying to get a wire out, couldn't get it out Friday have to have an all hands meeting and say, I don't know if you're going to get paid. And, and meanwhile, the they're trying to get loans and backstop, you know, backstop funds. And then tonight, they get the message, okay, your deposits will be valued and whole. So it's a rollercoaster ride, every founder signs up for the roller coaster, we just didn't know, we were signing up for this roller coaster. And in particular, by the way, last point, you know, as an ad tech podcast, really, there's some actually pretty profound impact for edtech. In particular, many of our clients, our school districts are universities that are federally funded, there's a way in which our, like, our financial solvency and potential liquidity is, is actually really, really important for us to secure these contracts. And so we were looking at people being in breach of their contracts, because they would have had to report or disclose their like financial situation. So I do think that it's an interesting set of conundrums edtech founders were facing fortunately, much of that has been averted. I'd love to just build on that. Because I think a lot of people spent this entire weekend kind of contingency planning. And, you know, I think there was some kind of Twitter chatter about like, you know, well, the whole reason for the 250k. backstop is, once you have that much money, you know, you're supposed to be kind of financially sophisticated enough that you would avoid a problem like this. And I like, sort of understand that, but and then you get the people who are like, Well, why would you just have a million dollars sitting in a bank account? You know, I think what's important to think about here is, if you are running payroll for a 15 person company, you have more than 250k, sitting in a checking account, like you cannot run payroll, for a 15 person company, without having more than that amount of money in a bank account. You literally cannot. So, I think there's this like kind of dissonance, where it's like, well, you know, this is a problem for the rich, and it's like, well, actually, this was payroll, like, a lot of companies did not pay payroll on Friday, because of what happened here. And a lot of companies were not going to be able to pay payroll next week. You know, even if the depositors were made whole, at some point in the future, like at some point in the future was not a acceptable, or was not a fair outcome, because it's like, you're gonna miss payroll, and you're gonna miss you know, your, you know, ability to pay bills and that type of thing. And exactly to your point, and, you know, you are going to have to report effectively insolvency to your customers. And so I think that's the ripple effect that the FDIC and other folks kind of involved here are trying to protect against is it has nothing to do with the actual like shareholders of Sv. They've, you know, they're done and dusted. Nobody's nobody's getting any more money from them as equity holders, but I think like the ripple effect of, you know, at least 100,000 companies not being able to hit payroll and not being able to pay their bills next week. There's there's just a crazy butterfly effect from that. Yeah. And a lot of political fallout. I mean, I don't know if you notice some of the, you know, imagine you're the Biden administration. And you know, that if the federal government is seen on any level to be bailing out Silicon Valley, yes, you could make that case that you just made that it's about payroll, but even payroll is this is payroll to Silicon Valley errs and design enters, most of the country doesn't have a lot of sympathy for that. And, you know, Matt Gates came out immediately saying, Oh, let's see if let's see what happens. If they try to bail them out, we're gonna have a field day, because then they would have. And I mean, I'll tell you, Alex, even even given the situation, they're going to claim that it's a bailout. Sure. And actually, this is, so this is where I'd love to head like. So now that we have the Feds stepping in and backstopping the deposits, they're doing an auction. The idea is that eventually someone, will someone or a set of institutions will buy off the SVB assets. You know, what's the takeaway? What what are the what are the things look like going forward? And you know, what is the, what are you sitting with, as you think about the next few weeks, months and years? Based on the what happened here? I'd love to hear your answer that first, I gotta, I gotta get my hair. Sure. Well, I so there's, there's probably a couple of different directions, you could go with it. I saw one funny article today. That's like, Jonas Kupo. I'm not saying that, right? You're the Greek freak, the basketball player is so smart, because he has 250k stored in 50 different bank accounts. And it's like, that's, that's one solution. That's one direction we could go from Greece, that's probably what you exactly. And you know, that he that is kind of a product of the financial system that he grew up in. And I think it really speaks to like an existential question of like, you know, one of the reasons that the US has gotten to the place in the world that it's gotten is because of the rule of law and the institutions that we have that help facilitate, you know, all the amazing kind of tech and lifestyle stuff that we have not to get all kind of philosophical on y'all. But, you know, I think it's really important to think about, and, you know, it gets at the future of like, should we as a country have regional banks? Or should we just, you know, put it all into the too big to fail banks? And so I think, like, that's a bigger question than I am equipped to solve the answer. But I think there's going to be some real soul searching among the US financial institutions. And I think that's exactly why, you know, one of the main reasons why Janet Yellen of the federal government is stepping in so fast and saying, you know, whether you call it a bailout or not, let's try not to, but we're, we're gonna make sure that these that the deposits are insured, or you know, that they're that they're, they're covered, I think it's because they're trying to avoid financial contagion among the entire regional bank system, because it's Silicon Valley. And it was the 16th biggest bank in the US as, as we've all seen, and part of the reason for that is, it's a regional bank, it's a regional bank in an incredible, you know, dead center of one of the fastest growing industries in the world for the last 20 years. So like, of course, it's, you know, it's as great, but there are regional banks all over the country. And if they were all to, you know, people were to say, Oh, wait, I didn't realize that I you know, that this was so risky that a bank could fail, and day and I get home from work, and my bank might have failed, I better put my money on, put it in Bank of America or whatever, the big banks, you're mentioning that. I think that you know, I don't think anybody except Bank of America, and Citi and those folks, you know, would want that future. I think, you know, one aspect of this that I haven't heard a lot of people talk about, but I'm really curious, but you know, there's this letter come out from a, it's never been signed by almost 600 Different venture capital firms basically trying to put, you know, it's a statement of support for Silicon Valley Bank. It's signed by, you know, all the big, big names in Silicon Valley, the Kleiner, Perkins, the sequoias as well as some of the tech specific ones like, like owl and MGSV. But realistically, you know, Silicon Valley Bank is gone. I mean, they can support it, but it's gone as of yet, you know, this week, that's a lot of money that if it does go back into all of these things, and VCs needs to go somewhere. And I don't think they're going to do the the Greek you know, that Greek model of putting a quarter million dollars in many banks, but they're gonna have to put it somewhere in this whole lot of money. I wonder what's going to happen with that? I mean, are they going out on that front? I do think that there might be some financial mechanisms that people you know, turn to which are these insured sweeps. The idea being that you use you know, your core bank as the quarterback of your money, and then it sweeps it out and spreads it out to other banks. You know, my a lot of what you're hitting on like the three main The takeaways I had one was, the rate of change in our society is just very, very evident in everything going on with AI with politics. And now with banking, you really have to be on top of your things when a bank can go from, you know, huge market cap to worth nothing in 24 to 48 hours. And I think that that is a kind of risk of modern times. Matt, you mentioned kind of the digital facility in which we're able to move money around and create accounts and all that. So, so good for so many reasons. But we are in high velocity moment in history. The second takeaway I had when I was watching all this is, man, wouldn't the crypto people be crowing if FTX had not just melted down? There was another like crypto lender. And now the stable coin is like being brought down? Because it was with with, with the bank? Yes. So, you know, this was supposedly like crypto was supposed to be the hedge against these large consolidated bank failures. And in some ways, I feel like this is the nail in the coffin, because with FTX, it totally blew up, and no one was made whole. And they're still going through the bankruptcy process of, you know, distributing out whatever was leftover, whereas this one, yes, it failed within 48 hours, and within 72 hours, depositors have been made whole. And then I think the third thing is, we have to acknowledge that there's other single points of failure for edtech. And, you know, I was thinking about Amazon Web Services. Imagine if AWS just went down and totally crashed. There's no contingency plan for 50% of edtech companies. And so there is a way whether that's a cyber attack, whether that's insolvency on their part with. There's ways in which we've kind of aggregated on different platforms, whether that's a financial platform or tech platform. And I do think that it's important that we step back writ large, as an edtech ecosystem, say, Hey, how are we hedging our bets? And of course, like, Google Cloud has been selling on the Hey, you should be multicloud posted for a long, long time, I hope, I hope people at least take the moment to question, maybe that is a good idea. Maybe we're at a certain scale, where, you know, multi hosting is a good idea. So there's a number of other companies that have become like, so courts and the infrastructure of the innovation economy and of edtech. And, you know, it's, it's a good chance for founders to reevaluate and hopefully, you know, by Monday, as they say, you know, we wouldn't have missed a beat and payrolls coming through. So we learned the lesson, maybe the easier way. The one other point of failure that I heard talked about this week, I thought was very interesting was companies like Stripe was payment providers, because some of those are startups. And, you know, every b2c ad tech company, or a whole bunch of them would be in a lot of trouble if stripe, you know, went down. It is a really, I love your point about the velocity. And I'd love to hear you're talking about this to Matt, because I know you follow this, you're the speed of it so closely. I mean, one thing that this makes me think of, and it reminds me a little of 2008, which is you know, not what you're supposed to say when something is risky like this, but is in 2008, there was a feeling of, okay, there's, these finance guys probably know what they're doing, don't they? These guys on Wall Street, they probably know what they're doing. They're making a whole lot of money. And then, you know, the wheels come off. And you see the heads of all of these companies being like, we don't even know what these things are. We don't know what a subprime mortgage like hardly is we don't know what a you know, these derivatives things are like, and I think there was this, like feeling I mean, Occupy Wall Street came right out of this of like, oh, wait, these people don't even know what they're doing. And I think there's a moment I look at this list of 600 VCs who are all at the same bank. I'm like, I'm not sure they are doing like. I mean, it's it's a pretty freaky moment to see that like an industry like technology, like almost 50% of startups in the US Bank with SBB. What is whatever it that's crazy, talk about a single point of failure. Like it's, it's, it's a thing and I really, I wonder coming out of this, if there might be a little bit of a sort of, I don't know, revolutionary zeal, like the Occupy movement where people say, Why do we think these these guys on Sandhill Road are the only people who understand the world and tech like maybe They are kind of just, you know, following each other around in circles. And maybe it should be Bangkok that that, you know, maybe it should be Columbus, Ohio, maybe it should be Texas like, I think there might be this might be like, a nail in the coffin coffin of the Silicon Valley myth, which has been, you know, ascendant for a long time. Yeah, I mean, I think there's kind of two things that I spend a lot of time thinking about. And the first is sort of along the same lines of what you're saying, Alex is, I'm going to call it that the cockroach theory of startup building. And it's like, you know, there's the stat like, 90% of all startups fail, and the reality is actually probably even higher. And, you know, actually, to make this even funnier one of the few listen to the how I built this with Guy Raz, is he always he always he has two questions that every interviewer has. One is, were you a good student in high school? And the other is, you know, how much do you attribute your success to luck versus skill? And, you know, I think it's like, well, reflecting on the startups, the startup class of like, 2010 to 2020. It's like, well, you faced neither a giant bank run nor a global pandemic, like you were pretty darn lucky. Like, right. So there definitely was some skill, but the fact that you didn't have to navigate these like global calamities, made it, you know, one of the best times probably ever to build a startup. And so I think there's this element of like, just figuring out how to survive in startup land. And the the flip side of that coin is like, well, if I'm a founder, particularly if I'm an ad tech founder, how much of my time should I be spending? Thinking about my like bank sweeps? Like, depends, right? Like, there will be a company that will just like, you'll pay $5 a month, and they'll do they'll spread your money across, you know, 100 different bank accounts. But like, you know, I think what the reality is, is in while there's such a concentration here, it's like, well, you know, these guys are pretty nice to us. And it's really not worth our time to be thinking about bank risk. Because like the probability of success is so low already. If I spend any ounce of my energy thinking about where my money's parked, it means I'm not focused on my problem. And you know, whether that's good or not, we can debate but I think that's the theory. And do you think that theory is reinforced now that their products are safe? I have no idea that I mean, that is, you know, the question is like, by this, you know, effective bailout, whatever you want to call it, like, did we learn our lesson or not? You know, and the answer is, I don't know. Like, if, if the answer is like, I pay a startup five bucks a month to put my money into 100 different bank accounts? Like, it seems a little silly, but like, if it means I don't have to worry about it, like I would pay that money. Yeah, that's, that's, that's probably the tax that we ended up paying, is something like that. But yeah, I thought you were where you're going with the cockroach. One is that ed tech startups never die. There is a degree. Say, we have a resilient bunch in ad tech. And I've seen founders go who, you know, with revenue, plummeting, and you know, sales struggling and velocity, and they still squeeze 234 years out of it. So there is a way to where I was buoyed by stories of founders being like, well, I didn't see this one coming. But on Monday, I guess we'll figure it out. I mean, I was texting with two or three, who had their personal money in bank accounts in Wellesley, Massachusetts, of all things. were, like, Silicon Valley Bank had acquired their personal banking bank. And then meanwhile, their money for their startup was at Silicon Valley Bank. And it was just a total, you know, shit show part. I know, we'll like bleep that in the edited version. But it is just amazing how resilient our founders were and how many people were just coming out of the woodwork to help each other. So in a weird way, it kind of was like backs against the wall, and the ed tech crew just came through. So we'll have more on this topic. You know, coming up through the week, I'm sure. But I, you we should probably give our listeners a little bit of the, of the upside of where things are headed with AI and edu. Lots going on in that space. So with that transition, I don't know. Do you want to set the table for us, Alex? Sure. So you know, this. This news came came at the end of last week, but some other really interesting things happening in the EdTech world every listening to this podcast, we'll know that we've been talking AI and chat GBT a lot over the last few months because it is really a huge movement, especially for education. One of the things that came out this week was a leaked apparently leaked piece of information from a Microsoft that says that Chad GPT, four may come out as early as next week. That's the next version from 3.5, which is what's out now. And it may include videos, the ability to do AI videos, AI speech, in maybe like, some of the things we've been speculating about maybe a lot closer than they appear. That is a big news. There are some other AI stars, but I'd love to hear what both of you have to think about, about just the idea What if Chad GPG for comes out next week and has videos and has speech? And has all of these different multimedia assets? What might that mean for Ed Tech? That let me throw it to you. Sure. So I mean, I think the biggest question for me is syntax and analysis and aware what the delta is between, you know, three, 3.5, and four, in terms of being able to comprehend a broader set of materials and and turn it into a more subject specific argument. So what I've what I've argued with a lot of the kind of current examples is very good at giving you the answer that appeals to the most people, but it's not particularly good at giving you the kind of subject matter expert answer. And I think like, that is the leak that I'm kind of on the lookout for. Because as soon as tools like these can start doing analysis, like that really changes the approach you take to using them today. It's, you know, you take the base level, and it like speaks your language, but you still have to do a lot of the like tuning either. Literally like cutting editing lines, if you're writing, or you know, figuratively, if you are a software engineer kind of playing with the model. So I think that's what I'm on the lookout for. I think modalities are like, interesting. But that's that's the leap that I don't know if it'll be this one or the next one, or when but that's the one that I think is a real kind of game changer to what we do. Yeah, my, I would echo plus one, everything that said, and especially is going around the need for tuning for the API to be practically usable. And I actually don't think that that's going to change I, I think the in some ways, the larger the model, the more generalized the content actually is. There's this weird way in which actually the people I talk to who do the tuning, say, it's actually quite shocking how little data you need to effectively tune a large language model, but more data in a large language model. It's like, Just point 1% better or point 2% Better for most functions, I do think that the modality of video is going to be is probably the highest risk and highest opportunity. And, you know, I think this idea of personalized avatar, that's going to teach you everything, that's where a lot of people's heads go. But imagine like Canva like features where you could make a, you know, an infographic video, you know, just add a push of a button, I think that kind of stuff is going to be available like it next month. And that's pretty exciting. And, and could be really, really great for the world. I think there's just also a growing group of people, very concerned. And the folks who did the social dilemma, the movie about Facebook and social media and the down downsides of that they're doing a new round, called the AI dilemma. I read about this on the information. And they're talking about the ability to create deep fakes and their real ability to like impersonate reality. And already stories are coming out of, you know, elderly folks getting a phone call from what appears to be their kids and saying I need money. So there's just, I think what we're the Pandora's box has been opened. And what we're really seeing is, the more that you power up these large language models, the more diverse the use cases are, and that could be a good thing or a bad thing, but Matt's point about tuning. I think that that's the part that people are still missing is that the tuning is really the essential part. For the value, and without it, you actually get the wrong answer a fair amount of time or, or it's something that a good web search could have told you anyways. Yeah, amazing points, I think all around the circle here. I am trying to go as long as I can without sort of having that that backlash, you know, reaction, that sort of fearful reaction about AI, even though I totally agree that it you know, disinformation and deep fakes and all of this bizarre stuff, the ability to, you know, fake political scandals, or say that a real political scandal was faked by AI. I mean, you can do, there's so many things that can go wrong here. But I'm choosing actively to be a little bit, you know, optimistic and excited about the AI. Interview, Alex, I know I'm gonna it might be trying in the next few months, but I'm gonna try it might be hard. I think that from a product perspective, the fact that this might be that we could maybe be able to do video that quickly that soon or speech that soon is sort of triples down to me on the prediction that Ed Tech is going to be increasingly AI, you know, run under the hood, because of all the reasons, you know, we know about video, video is the main communication medium for most young people in the world right now. It the amount of online video watched by kids is astounding, like crazy, like many, many hours a day. And it enables even small players, like you're saying, Ben, you know, if you're doing an info graphic, I mean, think of Khan Academy, Khan Academy was one of the biggest edtech, you know, events in the last two decades. And all it was was a guy making videos in his closet literally, like you, anybody can do that. Now, anybody can do that now on almost any subject. And yes, the tuning the accuracy is obviously important. But that's a pretty big sea change to go from, like this one person is unique and all of humanity, and that he's going to sit and teach everybody through video in his room, you know, to oh, now every buddy, literally every single person can do that. And companies have to just build incredible stacks on top of that, to make the videos, engaging, entertaining, perfectly accurate, pedagogically sound relevant to whatever the kid is interested in. I mean, there's so much potential here, and I'm gonna, I'm gonna try to stay optimistic about it, I just think this accelerates. If it does truly come out, whether it's next week, or you know, three weeks from now, it doesn't really make a difference. But if there if, if we're gonna go from text to video, that fast, it's just hard to even wrap your mind around how fast ed tech might might change. I always use news ELA as an example here, because I think they're such an interesting one, you know, Newsela, has spent many years creating leveled content, differentiated content for 1000s and 1000s, and 1000s of articles, if an AI can learn how to do that, very quickly, you know, that is a both a very big threat to their business model, and maybe a huge accelerant to their business model. For all the reasons you just said, Matt, they're the ones who know how to tune they have the data set, they have the subject matter experts at hand, who know, you know, what it really means to be writing a different, you know, grade levels accuracy, accurately. And so, you know, there, and I think a lot of companies are going to face this moment, even maybe YouTube for education, where it's like, how, you know, we can either find a way for this technology to take us to a completely new level, or we can be eaten alive by it because by like three people in, you know, in high school, making a startup, pretty nuts. But if content is is essentially free to generate, and it essentially the quantity of content is unlimited now, really, it's going to come down to curation, trust selection. And there's a way in which like, the Khan brand is also a good analogy in this because YouTube has exploded since Khan first started, and yet people keep using Khan because it's trusted, and it's curated. I think one thing I'm wondering about is how does this change the workforce? And, you know, what does workforce learning need to look like for people to thrive in a place where AI can be your assistant, your extension, and also where the rate of change is just so incredibly fast? I don't know. Matt, you follow that space quite a bit how, you know, AI embedded into workforce and upskilling if feels like all of us actually eat like an AI upskilling course just to be alive in the next year. Yeah, I mean, just in like perpetuity like you have to spend an hour a week on your AI stuff. It's really a question and bear with me a little bit because I'm going to start with a story about children's learning. And search did an article this week that I really liked about the kind of third I think they're like third graders that are learning with Amira and soap box lab and ello, and some of the other kind of voice recognition tools that are coming into the particularly that their early elementary population. And as I'm reading the article, I was like, Man, that must be weird that they're learning with like a, you know, a, an AI bot. And so that was my for like, the first 15 minutes I thought about it. And then, as I thought about a little deeper, I went back to I learned how to type on an Alpha Smart. And if you guys know what alpha smarts are, like, a very specific how to had a brief moment in the 90s, where that was like a great 90s and 2000s, where it was, like how you learned to type. And I remember bringing bringing mine home, and my mom and dad being like, What a weird device like, like it without getting too deep into it. But they had a similar reaction. And, you know, I think if you look at my parents generation, a lot of them are like the hunt and peck typers I think most people have a visual for that type of paper. And, you know, they're extremely smart people, and they succeeded in the in the workforce, you know, without the Not My parents specifically, but like, you know, society existed, that was fine. But you know, think about the like, step change of being able to type in general 100 words per minute that most kind of millennials and below have. And now it's like, okay, well, now you're whatever generation you know, today's third graders are, they're going to be trained to use voice and thought as their kind of modality for learning and for interacting with the world. That's wild. So I think it's incumbent upon us, you know, old fogies here to learn how to adapt. Right. And you know, your reaction, Alex, I think makes a lot of sense. And I think is an easy one to kind of default into. But ultimately, the folks who are going to succeed in the workforce, and finally getting around to your question, then are the ones that will say, like, well, I, you know, I can go hide, or I can dip my head in and figure out how to use these things. And it's, it's super easy to say like, of course, I can go learn how to do it's very different to like, actually go and commit the time to learning them. They're, they're simple tools, but they're not there. They're hard to use Well, and that's what I'm looking for. Yeah, and I will say, just in the in the news, we saw people addressing different elements of the spectrum that you just mentioned, we've had Salah from fourth grade on the show before. And, you know, she really is actually focusing on the upskilling of, you know, tech talent to incorporate artificial intelligence, but all the way down to that kind of elementary use case, AI. Edu, a nonprofit, just did a call national call for educational standards around AI. And they had 50 plus of some of the best known nonprofit and for profit companies joining their pledge, which was the idea that it's not enough to have computer science as a requirement for high school that we really need to be teaching the fundamental building blocks of AI all the way down in elementary. And that's not because everyone's going to be an AI programmer, in fact, like, none of us are going to be aI programmers, right? Because they will help us program. But it's the idea that you need to be metacognitive, about what these tools are, how they work, how they you know, what is the answer on giving given from the AI? And how might that answer have come to be so that I can evaluate whether it's accurate or not. And then Salesforce also released digital skills now survey doing that only one in 10 global workers have in demand AI skills. I mean, the reality one in 10, I thought, Man, that's pretty good, like 10%. But I think if you dig a little bit deeper, it's just the ability to use tools that leverage AI, not even to build AI itself. And we also saw on the news co.org, their CEO talking about the impact of AI on teaching computer science. There's almost A whole rethink of what needs to be learned, what what should people be trained on? And I think, you know, the, these headlines tell us that it's, you know, for, you know, my son who's six, he's going to be an AI native, just like, you know, other kids, you know, the current generation, they're very mobile phone, mobile natives. There's a way in which this is going to kind of feel like second nature. And so it does represent a big shift in both ad tech tools. As Alex, you and I, we talked about all the time, but also in pedagogy. Oh, yeah. I just think it can't be understated. The this the code.org CEO, interview is really interesting, because, you know, it took a long time between when coding was invented, for people to start saying, hey, maybe we should be teaching this in K 12. School and training teachers on it. And even now, I mean, it's, it's beginning to happen, some states have it. But yeah, in places like code.org exist, but it's still kind of struggling to be truly put on the same level as as reading, writing and arithmetic. And yet AI the way we are considering it has been around for just, you know, I mean, AI was invented 50 years ago. But you know, the way we're thinking about AI right now, it's that much more recent, and there's already this humongous movement to incorporate it into K 12. I mean, the kinds of partners you're talking about Ben, Google, Microsoft, Intel, Nvidia, at&t, Hewlett Packard, GitHub, Teach for America, Boys and Girls Club. I mean, this is like, this is not small, small, you know, folks, GSV TechStars, Penn GSE, like, right off the bat, the entire education and tech ecosystem. Tech, even by itself is saying, Yeah, we really need to get ahead of this, because there are a lot of risks, but it's also going to be the new literacy. I mean, quick story really fast. When I was in high school, I had a public access TV show with a few friends of mine. And imagine you're like Wayne, or are you like, Garth? Pretty much? Yes. I spent a lot of time behind the camera, actually. But we, we always wanted to know what people thought of the show. And you know how we had to figure that out by putting our email address on paper and showing it on the video on the television? because how else do you know, in that era, who's watching you don't have likes? You don't have views? You don't have? You know, comments? You have nothing. And that was how, you know, that was how we did it. We did a live show so people can call in, just so we can understand that there's anybody out there watching. Now fast forward, your son, then? I mean, can you even imagine that if he makes a video, it's immediately, you know, everybody in the world can see it, anybody in this class can see it, he can get 100,000 views in, you know, three days. I mean, that is a different world. And this AI is is the next phase of that. I mean, the same way that a YouTube and a Tiktok are basically curatorial systems. They're like, yeah, we're putting out more content than anybody's ever imagined all the time. And it's up to you guys to figure out what you want to watch. I mean, that will happen with AI, people will be creating videos of every kind to teach everything and then not only include traditional educators, but workforce, I mean, as we're just saying, I mean, we're going to be in a world not that long from now, where any of us at work, no matter what we do, we'll be able to, you know, hit the Siri button and be like, I need to know how to do this, and it will auto generate these not very, not Siri anything but Siri. Okay, sir. Yeah, open AI, whatever the name of the future AI overlord robot will be, but you know, it will say, hey, teach me this thing. And it'll just teach it to you. It'll, it'll teach it to you, whatever, however you want, he'll make you an article and make you a video. And then kids will grow up like that. And they'll think, well, you know, that's what education is you decide what you want to learn. And you tell the system and it teaches it to you. It's like a really nuts. World. I just add to it. Just to add to that, Alex, I was pitch on Friday, by a startup called learned on XYZ. And they will make you instantly a like two hour course, on any topic you want. And you just input what I want to learn. And it creates a multi module course, with videos with readings with activities and assessments, and it takes about 60 seconds for it to generate an entire course. Yeah, I mean, that's incredible that it already exists, but it's like you That's where we're going for sure. It's the fact that it's already here means it's the sky's the limit. I mean, that's amazing. And I mean, I don't know, I just think we can't even begin to understate or overstate how different it's going to be what education is going to look like in the future. When tools like that exist. I mean, it's just changes entirely. What it means to educate yourself is as YouTube and Wikipedia already have, but you know, it's just a whole different ballgame. It's really not or tick tock already has, I mean, we talked last week about that survey that came out that were more than half of college students said they learn more from tick tock than from their classes. Did you guys see that thing? So like we're in a really new world here. There are other AI news that came out and Matt I want to bring you in on this was that grammerly expanded so you know we've seen a bunch of turn it in and a bunch of different tools start bringing AI and we saw notion bring it in this week. Miro bring it in. Grammarly extends beyond proofreading with AI powered writing. So that's really interesting because Grammarly has for years now been a tool that's been a go to, especially for students or adults with learning differences or who you know, for any reason are not confident in their own writing ability. Grammarly sort of steps in and fixes everything makes it really, really clean. Now they're going even further and saying, you know, hey, tell us what you want to be writing about. Tell us what you want to communicate, and we'll we'll write it for you. So you're just seeing these things come every week, and all of them, I think will have effects on education. Um, Matt, I want to ask you about something I know we're running a little long here. You broke some news in ed tech thoughts this week about Google? Acquiring Photomath? Last February, February of 2022. Can you please explain that to us? Because I don't understand it. Yeah, so and, you know, I just want to like reinforce I think I've spending a lot of time thinking about all the questions that you just brought up. So thank you. On Photomath, the way this news came out was Google is being sued in antitrust court in Europe. And as part of the like discovery process for that any trust review. For for folks who are less familiar with like antitrust rules, and lawsuits, slight discovery is the lawyers for both sides get to ask for documents related to the issue at hand. And when it's as broad as like, does Google have competitive a competitive monopoly over the search market, they can ask for basically anything. So a lot of these kinds of interesting things pop up a Twitter account you might enjoy, if you're like, into this type of stuff is internal tech emails, I think it's at internal tech emails. And it's literally just a bunch of like, Discovery docks, that those are the ones that made it to trial. So the cases do have to go to trial for discovery to go public. And that's often the reason you'll see companies settle out of court is specifically so that docks like these don't get out into the public's hands. But so internal check emails that there's like decades of these great emails that that go back and forth, kind of like Steve Jobs, and Mark Zuckerberg and all these other great ones. But so this acquisition came out as part of the discovery process. And the first place I read it was this like random correlation. Like, you know, town newspaper that had some sort of a vested interest in knowing about Photomath. You know, there are also some more legitimate sources that I think made sure that the T's were crossed, and the i's are dotted that did in fact happen. But yeah, that's it's it's kind of interesting, how long they were able to keep it bottled up. You know, there's another one this summer where Google acquired bright bytes. And it's like, well, the founder of bright bytes, changed his LinkedIn description to like, Head of Product edu product at Google. Oh, yeah. Probably got acquired. I had not checked the Photomath teams, Lincoln's recently. So it's possible they to do, but that's another way that this stuff often breaks out. Yeah, I mean, I'm often on my soapbox here talking about Google. It's the largest education company in the world. The interesting thing about Photomath is it actually started I believe, it's bank software, where you know how when you or depositing a mobile check, it can read your cheque, and then you can deposit it. And so Damir, the CEO had actually sold a company that did, you know, visual recognition around reading your bank statements, and he was working on one for more sophisticated financials. And he just realized, wow, this could really work on my daughter's math homework. And from there kind of started Photomath. They were at a point where they were getting something like 100 million users. And, you know, annual users, they had a really foundational challenge, which was, everybody wanted to use it for free. And you know where to put the paywall. And so I know they've been working very hard to build in tutoring, so that you don't just use it to get the answer. But also, you can understand that three or four different ways to solve that same equation. But I think it's really clear now that this use of AI computer vision, plus, you know, AI, driven instructional units that can break, essentially, automatically create a lesson or unit around any problem was of strategic real strategic value to somebody like Google. And I do think that it should raise big questions. A writ large, around, why is Google buying up these, you know, really, really prevalent learning tools, and not disclosing it? Because meanwhile, you know, check is check. And some of these other companies have launched similar products, and they're labeled as cheating tools. Right. And they face all of this like headwind around, you know, Is it cheating? Or is it tutoring? Or is it instructional. And meanwhile, we've seen Photomath and the Google team keep a really low profile. So in one sense, this could be a really good thing for learners as Photomath potentially gets incorporated across the broad set of Google education tools. In another sense, this is potentially another market distortion that edtech entrepreneurs are going to have to deal with. Because now the largest edtech company in the world, Google, which happens to be the largest one of the largest companies in the world. Google Now also has the largest, you know, photo answer system in the world. Photomath. And, you know, if you're trying to figure out how do I make a business case against something, I don't believe Photomath ever figured out the paywall conundrum. So it's, you know, on top of all of this, you know, the full stack of Google offerings, aside from their cloud infrastructure, and Chromebooks are essentially free to the education ecosystem. So an interesting moment in time, and something that we'll have to watch, you know, per our prior conversation, the rate of change in AI, is accelerating so much, will a Photomath acquired by Google have an insurmountable lead by virtue of its position with major LLM AI provider lambda and bought and so on? Or will, will they actually not be able to keep the leaf given the rate of change and experimentation that we're seeing, and as you say, Alex's Cambrian explosion of, you know, AI, ed tech startups? So this one is one that keep coming back to and watching? It's Yeah, it's amazing. And you know, the only thing I would, the only thing I can possibly add to that is like Photomath is unbelievably popular. I mean, it has 300 million downloads, has more than 200 million ratings on the Android Google Play Store, which is ratings are usually a pretty good proxy for how popular something is. Photomath is everywhere. And it really caught like wildfire. You know, when it when it launched, the idea that Google sort of instill a top five app, I think it's the number one math app in the world and top five education app, the fact that Google, for one thing, they own the Google Play store, they know exactly how many downloads they got, but they can sort of gently behind the scenes, like you're saying that and been, you know, swoop up this app and say, you know, this is ours now and we'll figure out what to do with it. At some point. We'll incorporate it, maybe we put it into Google Suite, maybe we just put it into our own future AI plans, because it's also, you know, using artificial intelligence. And it's just it's a very interesting moment. And I know Ben, you're you're always justified your soapbox about Google being the the big behemoth in the room. It just we keep hearing things and make it more and more true. Last subject really fast. I was not sadly at TSA. up by Southwest edu This year, I wish I could have been there. I had huge FOMO I really wish I could have been there. It sounds like it was an amazing event. The two of you were both there in Austin, living it up talking to everybody networking. Tell us what South by Southwest edu was like this year? Then let me let me kick it to you just just start that off. Yeah, I mean, it's amazing how last week feels like two different weeks, there was the SVB week, and then there was the South by Southwest week. So, you know, for those of you who don't know, South by Southwest, is this popular music and cultural festival, but they also have an edu component focused on early childhood, k 12. and higher ed, not as much corporate work learning focus there. So tends to be, you know, aligned with academic institutions. And first, we had a great a tech insiders happy hour, over 200 of our best friends, including Matt showed up at the seven Grand Bar in Austin, it really was a blast. And it was just a reminder of how strong our community is coming out of COVID coming out of Ed Tech winter, the optimism in the room, the kind of shared problem solving, it was it was a celebratory moment, it really was great. I'd say the other three big takeaways from South by one is that most of the school districts were not present. If you have a teacher shortage, if you're having budget concerns, you do not send a delegation to South by Southwest, and they were noticeably absent in in this time. And I do think that the kind of presence of K 12 institutions and university institutions is declining at conferences overall, just because the staffing shortages and budgets have been tight. And the second is that there was a lot around AI use cases. And I think it's still, you know, early days, in terms of how people are thinking about where a AI fits into that go for it ecosystem, in part because it's changing so fast. And in part, because many of these big institutions aren't willing to do a deal with somebody, they're like, you know, it's like four dudes in a garage. Like there's a way in which you have to have a certain gravitas a certain size and a certain stability for big players to adopt you. And so, you know, we might have a meeting, you know, happy hour out in the bay area where everybody's really excited about what's happening, I think it's a little bit more of a cautious like, stay the course tone, when you start hitting the middle of the country. And they are thinking about much more, the larger players. And then I would say the third big topic was really around the mental health crisis. several sessions just pointing out that at all levels of our system might be students, it might be educators might be parents might be professors, this idea that the education point of sale is also the mental health point of sale, that the idea that health care and learning are kind of converging as both integral to you know, the person's outcomes in life, but also the idea that we need to be more synergized, and how we deliver those things that was really salient across both the networking and the sessions. And I found myself meeting with health care people who found themselves in an education conference. And that was really an interesting confluence. So I do think that that's actually something we're going to see more of. So maybe as the schools and universities back out, we'll see more health care companies, and, and nonprofits and so on thinking about, you know, let's meet our constituents where they are, which is largely in schools and universities. Matt, what were some of your takeaways? Yeah, I think just on that last note, the Community Schools movement that has I think, is mostly taken written California and is starting to spread a little bit across the country is something I following in general, and it's cool to see it kind of live and action at South by you know, and I always kind of get a kick out of looking at the sponsors for these conferences and seeing like, you know, what, they they do it in this kind of very capitalistic leveling system, or like you've got the premium and then the like, kind of premium and then like a little bit premium, but more not so premium, and it just goes down the stack. So I do encourage folks to take a look at those because I think it shows kind of, particularly among the ecosystem players like where they're prioritizing their dollars across different conferences. For me, I I think my takeaway was kind of it felt a little bit more intimate, in that, you know, not everybody was there. But there was a good time for like one on ones a lot of folks trying to meet kind of individually, we have the happy hours, and I thought you all really kicked off the week strong. With with your happy hour on Monday, I thought that was a good way to like, kick things off at the bang. But but for me, it really was a story around getting together with folks. And, you know, the one thing that was a little surprising is there wasn't like a major company announcement. So at GSP, which is an April, there's usually somebody makes some interesting announcement, or there's kind of a cacophony of announcements around it. I think the to you edX deal was around GSP, a couple of years ago, but there's usually like a couple of big announcements, and I didn't I don't think I saw one around South By. So it might just be folks, folks. We're kind of waiting for mid April. But that's something that I tried to keep an eye out for. Yeah, really interesting. It's true. We didn't You didn't talk about any giant news that dropped last week at the conference on this on this, it turned out that it dropped on Thursday, Friday. dropped on on the entire the entire ecosystem. So if you're listening to this podcast, we're putting this one out, live to try to get it or not live, but you know, early to try to get it into your hands fast because the Silicon Valley stuff is moving quickly. That's why you didn't hear the theme song at the beginning. It also means that the we're going to take the m&a and funding and put it in the newsletter rather than listing it out. But there's actually a lot of good funding rounds this week, coming out of Europe, especially. But it's been an awesome episode. I know we went a little long because we didn't edit. But it's you guys are amazing. As always, man, I think you're like the investigative journalist of ed tech that I feel like you're digging through the internal emails to find the real story. It's really interesting that Woodward and Bernstein and Ben as always, you know, amazing takes on everything is an amazing moment for this for this world for the tech world. And the Ed Tech, you know, announced on that point, too, I just think it's I'm feeling especially grateful for the Ed Tech community with everything that happened with SVB people were standing together, they're supporting each other. And with everything going on with AI and this incredible ride we're on. I think we have a lot of great people leaning in trying to do what's best for our customers, our learners. And for the society as a whole. It's going to be a really, really interesting turning of the corner this spring into the rest of the year. I don't think we've ever experienced something like what we're about to experience in ed tech land. And I'm so glad to be processing it with the two of you. Yeah, and we're gonna see the Occupy Sand Hill Road movement starting there we go. I'll see you out with the I'll bring the beer that sign. Camping gear. Okay. All right. Thanks for having me. Yeah. If it happens in ed tech, it's gonna you're gonna hear about it here in ed tech insiders.