Brian Kardell and Eric Meyer chat with Microsoft's Clint Covington and Peter Wu about the many integrations of Math, the role MathML and some amazing statistics.
Eric Meyer: Hello and welcome to Igalia Chats. I'm Eric Meyer and I'm a developer advocate here at Igalia.
Brian Kardell: And I am Brian Kardell. I am also a developer advocate at Igalia. And today, we have two very special guests on the show, first time guests from Microsoft. Guests, do you want to introduce yourselves?
Clint Covington: I'm Clint Covington. I lead the productivity effort or the accessibility effort for Microsoft productivity products. So, that means Word Excel, PowerPoint, Outlook, SharePoint, Teams, OneDrive, most of the things that people use for daily work.
Peter Wu: Hi, I'm Peter Wu. I'm a software engineer at Microsoft. I work on PowerPoint and in the past over a decade, I've been focused on the accessibility of the experience and helping people create accessible content using PowerPoint.
Brian Kardell: So, you might be wondering, what does that have to do with Igalia? Why are we having the Microsoft people on Igalia Chats talk about Office and PowerPoint and those kinds of things? And it is because of the intersection of the accessibility work and specifically some interesting stuff that Microsoft does and actually has driven for a long time about math. And yeah, I don't know if you want to talk about any of that. Clint, you had some thoughts on it being a sort of a bigger effort.
Clint Covington: Yeah. So, things at Microsoft have been really interesting for the last couple of years. About two and a half years ago, we had some employees approach our executives and say, 'Hey, it's really important for us to be able to use Windows out of the box for daily work.' And our executives came and said, 'We really want to support this and we want to make sure that we've invested to make sure that Windows has a great out of the box accessibility experience.' We determined one of the big gaps for that was the screen reader experience. So, we started funding work to improve narrator. We've made, I don't know, 150 bug fixes at least. We've added new features such as a BrailleViewer and Braille HID support. We've been working closely with the M35 apps teams like Word, Excel and PowerPoint to build a great out of the box experience and just go through the freeway scenarios and tick those things off. So, we made improvements to spelling and comments and the PowerPoint design view experience. We're working on the PowerPoint slideshow experience. And PDF, I know you guys care a lot about PDF as well. And so, that's an area where we've made a lot of investments on. One of the areas that came up in there was math. And we had this realization I think that math, especially inside of Word, is particularly important to the blind community.
Brian Kardell: Yeah, definitely. Actually, I watched a talk that I know Pete had been preparing this talk for a while. Y'all did some work with Daisy, is that right?
Clint Covington: Yeah. It's really interesting. Richard Orme, I don't know if y'all know him, but he's the CEO of Daisy. Daisy Consortium is a group of organizations that care deeply. It's a nonprofit that care deeply about accessible content. Well, Richard reached out to me and said, 'Hey, Clint, we would love to work closer with you on improving the math experience. We have a math group of people, experts in the community that care a lot about math and there's some bugs, there's some challenges.' And when he made that call, I'm like, 'Oh crap, we don't really have anybody working on the math stuff right now. We got to figure out how we're going to do that.' But the thing that Richard really highlighted to me is if you're blind and you need to do your math homework, you're not going to use a piece of paper and pencil. You're most likely going to use the word equation editor because it really is the best-in-class for that experience. And what that flipped for me in my head is up until this point, I always though of equations as just like another one of the content side of work. You have tables and lists and comments. Equations is just one of the other things. I think Richard's conversation with me helped me realize that the equation editor is more of an assistive technology for that community. And when we look at assistive technology, we stop looking at MAU and how we're going to drive broad usage to things. We start to look at how important is this for that community and how blocked is that person from doing the thing they need to do? So, I think we realized that math becomes a gateway into STEM. And when we have friction in that math experience, we start to limit career opportunities for the community. And so, that caused us to go back and re-look at the math experience with a much deeper lens and say, okay, this is an area where we can't just be good. We have to be great. And we have to think even deeper and harder about the accessibility experience and really treat the equation editor more as an assistive technology and fund and resource it appropriately.
Brian Kardell: Yeah. So, Peter, you gave this amazing talk and I took so many screenshots of it and I did tweets and everything. Not tweets because I'm not on Twitter anymore, but social media posts. And yeah, I mean, I love some of the things that you pointed out in there. The number of equations and things that you were talking about. Can you just for listeners, give us some of those amazing high level things that will make people go, 'Oh, okay. Wow, I didn't really realize that. Now I see why we're talking about this.'
Peter Wu: Yeah, absolutely. I can start out just with a little bit of an introduction about how I got started working in math a little bit over a year ago. I had worked on it a little bit before then, just in general, trying to get all of the types of content in PowerPoint and other kinds of documents to make sure that they're accessible when you're reading them in the app and when you export them to PDF and things like that. And it was around April that Sara Shunkwiler, a instructional designer at Johns Hopkins, reached out to me on LinkedIn about how she was really struggling with making math accessible. And it was due to limitations in our tools and having to use several different tools in her workflow and really just wanted this process to be a whole lot easier. And I realized, first of all, I reached out to her to learn more about that and wanted to let her know about the accessibility of math that we do have built in to Microsoft 365 apps. And in doing my own deep dive into that to kind of explain it and walk through it, I realized that it wasn't working as well as it should really. And there were some shortcomings. And then through talking with her, I learned more about their workflow and about how things that they were expecting to work or wish they would work that way just weren't working the way they wanted to. And people were resorting to doing things like using add-ins and taking screenshots of math or typing up math in plain text. And those are all things we didn't think were a good experience for people who needed to consume the content with assistive technologies. So, I went and tried to understand what's wrong here and try to fix it. So, I wasn't an expert in the area at all, but I dove in and learned and started to make changes to PowerPoint and then Word and Excel and OneNote as well to improve the experiences there. And in the talk that you attended, I was giving kind of a summary of what's happened in the past year and what sort of impact that's had. The one that's very surprising to us is that bringing in math from other sources, in particular the web with MathML is a lot higher than we expected. What we did is enabled copy paste of HTML that contains MathML in it. And there's a lot of it on the web now.
Brian Kardell: Which you wouldn't think is that high.
Peter Wu: Yeah.
Brian Kardell: And even like I'm the chair of MathML group. And when I saw this, I was like, wow, that is higher than I thought. So, continue. Yeah.
Peter Wu: Absolutely. And thank you so much to Igalia to pull off this incredible piece of technology that the whole world will benefit from. Having native built-in MathML support in every major browser is a real game changer. And in the short time that that's been available already there is an enormous amount of math on the web. And so, we were seeing on the order of magnitude of like 150 million equations from HTML getting pasted into apps like PowerPoint and Word a month. So, that blew our mind and just shows how powerful that is. It overshadows the amount of math that gets created directly in the apps using our ribbon and things like that. And those all come in an accessible form. And so, we really love that. From there, you can read it with a screen reader in our apps and you can also publish it to PDF in an accessible form and consume it there. With certain PDF readers and screen reader combinations, you can read that math in a hierarchical way and really have a great experience understanding it. Yes, that's right. It's a little bit lower in the summer months, but that's about the level that we were seeing going into the summer. And then we'll track how it's doing in the fall. We expect the numbers higher as more people discover it. And more sources of content come online. And also we've also made some improvements that expand our coverage. So, it's really exciting to see that all come together and to see the usage really go up like that.
Clint Covington: The other day, Peter enabled paste that had LaTeX in it as an H&L attribute. And I just checked the scorecard and we like 2 million people in the last 30 days or 30 million equations in the last 30 days got paced. It's one of those just goes like the upside about how quickly people picked up. I've never seen something go from effectively zero to as much usage as we had almost overnight. And that's just because people already were pasting HTML. They were already trying to do the scenario, but what they were getting was gobbledygook. And when you think about accessible content, you're now students who are showing up in class and before they might have had... The professor would've replaced that with an image. That's what Sara and many folks were trying to do. They replaced that with an image, which isn't accessible to anybody or to excessive technology. So, we're excited about how it just makes that whole world of accessible content much easier, much more approachable for people so they don't have to do the extra work of recreating those equations inside of Word or PowerPoint. Yeah.
Peter Wu: I'll give you a little bit more context on this new growth of bringing in LaTeX that's in HTML webpages. What I had noticed is that there are some major websites out there that use KaTeX to embed math in the page. And KaTeX has, well, it typically starts with a LaTeX source content. And then you can configure it to render something that'll display in the browser as well as generate a MathML that can be used by assistive technologies in the browser. But what we found is that some websites configured tech to not include the MathML. So, that's unfortunate. But even when they do that, KaTeX will write the LaTeX content into the HTML markup in an attribute. And so, we just did a little customization to our HTML import to recognize, and it's very unambiguous. KaTeX has specific class names that it's using in a specific attribute where they put the LaTeX. So, we can kind of just scope it down to that case. We can find that LaTeX and import it. And that just has a huge impact. We found that there's actually quite a lot of webpages out there that are configured this way. And now we can bring that math into Word and PowerPoint and OneNote. And prior to that, the math was just skipped. Either that or you would get the raw LaTeX source. And from there, if you knew what you're doing, you could convert that to native math, but obviously it takes some work to do it.
Brian Kardell: This is interesting to me too, because you heard Eric kind of say, 'Wow,' but you didn't get his eyes bug out of his head when you gave that number. We can see on the video here. That's how I responded too. But they're not just mathematical jokes. These are important things that have to do with STEM education and maybe important research. This is a big societally important thing. And for better or for worse, the web is not the main way that STEM shares math. I mean, I think historically it's in PDF form. So, you also have done a bunch of stuff with PDF. And PDF, I think you said was, I have a screenshot here, 347 million equations a month from PDF. That is amazing. I mean, I would like if more of that came to the web, not in PDF form. But the fact that these all interoperate, you can copy and paste between them, they all work for accessibility. This is so great. I just think it's-
Clint Covington: Those numbers are so big. That's an equation for every human in the United States. Sometimes, we just see these big numbers and we're like, 'Oh, what does that mean?' There's another area here that I think is also interesting and that came from feedback we had from Daisy when they went out and talked to their users. They're like, 'Sometimes, we get confused as we're navigating through an equation. And we're really good at telling you when you leave a fraction or a numerator or a denominator, but we weren't telling you when we were entering that part.' And so, I don't know if you all know Murray Sargent. So, Murray's been working in math for decades. He retired a few years ago when I got the call.
Brian Kardell: He was an OG MathML guy from the late '90s, I think even. He was an early Microsoft employee, very early Microsoft employee.
Clint Covington: He was. And he's the one who wrote most of the math code inside of Office, especially the equation editor. So, Murray's like 84. He's still incredibly fit, super active. So, when I got this call from Igalia and they started telling me about... or not from Igalia, but from Daisy, one of the first things I did is I called up Murray and I said, 'Hey, Murray, we're getting some feedback on your math experience, but there's some bugs and challenges.' And he's like, 'Clint, I would love to come back and help and work on some of this stuff.' So, Murray came back and he's been helping Peter and myself ramp up in the area, but he also dived into this area around start equation, end equation. So, that as you're moving through, you'll hear start numerator and numerator. If you go into a square root, start square root. And so, it's been fun to have Murray come back as a part-time contractor and participate in helping us rethink our math strategy and point things out to Peter because Peter is fairly new in this area. But for Murray's take a moment for us to sit back and look at his contributions that he's made throughout his career. And recently it was a real privilege to present to Murray a lifetime achievement award from Microsoft for his work in digital math. And I think that's just part of us trying to recognize that the work we are doing is really building on the shoulders of a lot of other people. Not just Murray, but folks like Neil Soiffer who's doing work in MathCat and the work you all have done inside of the browser for rendering math. And so, it's just been fun to get to know many of the people that have been involved in math for a long time. When we share this impact, I think everybody's been pretty surprised at the growth that we've seen in math.
Brian Kardell: Yeah. We'll include a link to the article on the Microsoft 365 Insider's blog, Meet Murray Sargent: The Quiet Force Behind Accessible Math. I'm really pleased to see him get some of that recognition because he was a champion for all those things for a really long time. And not just any champion, but there were times when most of the world had kind of given up on MathML. We got it so far, but then there's this complicated history where it got taken out basically. And then it took a step back and overnight lost a billion potential users or something like that. And it just kind of went downhill for a long time and it got stuck until it came about. Igalia tried to rejuvenate it and get it through. And we got some funding, we got a grant, but we really tried to find people to help sponsor the work and it was really, really difficult. Even big publishers and browsers, nobody really wanted to invest in it because there's not a big return on investment you can point to. It's very difficult to explain to somebody why this is so important, especially people who write checks. And so, at one point, we were kind of stuck and Igalia was just footing the bill all along and it was getting really difficult. And we decided to start a collective. And Eric gave a great talk about how we could use collectives to fund a lot of important work and do important things so that no one organization or person or group of people has to foot the bill, but we fund it a lot more collectively. And to launch that effort, Neil and Murray provided I think $120,000 out of their own pockets to launch that effort and really helped us push it over the finish line.
Clint Covington: That's awesome. At the Ability Summit, it was one of the most moving experiences where we had a guy named Alex who's a blind developer. He works on the Excel team. And just as he entered college, he'd lost his sight. And he was a computer science major. And he was just like, 'I wouldn't have been able to get through college if it wasn't for the work that Murray had done in the equation editor inside of Word.' And so, Alex and Murray had this conversation where they talked about his experience and how well the code worked and how Alex reflected like, 'I would not be at Microsoft today as an engineer if it wasn't for the work that Murray's done.' And I think one of the things that from there that has been super insightful is we're not just talking about math is more accessible now for the next six months. The work we're doing here will live on probably past us being around and even involved in technology. And so, for me, I love to see us lay this really solid foundation that folks like yourself have built on, that Murray have built on, that allows us to put people in a place where they can have equal access to STEM education.
Peter Wu: In my experience being relatively new to this in the past year, hooking all this up and seeing the great impact that it has was that I know that the foundation that all this is built on was something that took multiple decades to put into place. And it takes quite an investment, quite a tenacity to stick with it that long to see it through, to build all of that that's needed. So, really appreciate everybody who's been involved in that effort and really pushed through. And I think that maybe throughout a lot of that time we weren't seeing the payoff, but now we do. It's amazing how all the pieces have come together now and we're seeing this accessible math at scale throughout the world. And we really want to change the equation. There's a lot of inaccessible math out there, but the mix is going to change to where a lot more of it is going to be accessible math and that's going to make a huge difference to everybody.
Clint Covington: Hey, do you want to talk a little bit about the work you've done in LaTeX and just improving the overall support we have for LaTeX? Because I think when you talk about STEM education, LaTeX is another really important part of that.
Peter Wu: Oh, sure. Yeah. So, as a grad student in the '90s, that's my first experience with LaTeX. And I thought at the time, well, this is neat and all, but personal computers and GUI was coming out and I liked apps like Microsoft Word in the very early days. I was thinking that's really going to take over because your average person is not so technical to program, write code to write a document. They can just point and click with Word. But here we are 40 years later and LaTeX is still going strong. So, there's a reason behind that. And so, I recently got back into that in talking with Sara Shunkwiler and one of her associates, Jeffrey Kwan from the Ohio State University that she works closely with. He's a deep mathematician and researcher. He writes tons of papers in LaTeX. And he said plainly, 'There's a whole community of people here that if you want them to use Word and PowerPoint and do math in them, you have to support LaTeX. End of story. They will not even start trying to use it unless it's there.' And then so I looked at what we had, which was starting in 2018 I believe we had LaTeX support in Word. And quickly I was able to hook that up in PowerPoint so that we can have it there as well. But then through the course of doing that, testing it out and trying it, and as I got onto these research papers that people were writing, I found, well, it can handle certain things, but there's a lot of things that it's going to trip up on. I found that in analyzing the LaTeX that's used on Math Stack Exchange, for example, it only handled about 90% of the keywords used there. So, we really needed to make it more robust if we wanted people from the LaTeX world to come in and use Word and PowerPoint. And I was just getting started with AI coding agents and seeing if that can help us accelerate our work. So, I gave that a try with just revamping LaTeX in Word and PowerPoint and such and built a whole new LaTeX converter and really put it to the test in terms of real world content and tried to just get as much coverage as I could so that people that are used to using it for their research papers and other materials like that could easily come in and make that work for them in Word and PowerPoint. So, we have a much, much more robust system now, like 99.5% keyword coverage of what's used on Math Stack Exchange, for example. Yeah. And I also found out in the course of looking at all that MathML that we were bringing in from the web, and I discovered as you were saying, no one writes MathML. All that MathML was auto generated and it was generated from LaTeX. So, basically there's around four major platforms that convert LaTeX to MathML, but the source is LaTeX. And having a direct line for that LaTeX to get into Word and PowerPoint was really important. So, we invested in building that.
Clint Covington: And then back the other way. So, we go back from LaTeX into MathML.
Peter Wu: Yeah. Absolutely.
Eric Meyer: Mm-hmm. Yeah. Yeah. I have experience with one of those four major... Well, maybe it's one of the four major platforms. I used MathJax for a project once. I actually have a copy here. This is the Effects of Nuclear Weapons Third Edition of 1977, which has a whole lot of equations in it. I can show, here's an example of pages of equations. Anyway, this was published by the Department of Defense and I studied nuclear policy for history and a friend and colleague is also big into nuclear stuff. So, he pinged me and said, 'Hey, I'm thinking about...' Because this was only ever available in PDF form. He said, 'I want to do it in HTML form. What do you think some of the structures should be?' And I gave him some advice and I said, 'If you want some help, let me know.' Well, yeah. So, we ended up doing a ton. I basically rewrote all of the equations as LaTeX, but then used MathJax to turn them into whatever. I have occasionally thought in the past year or so that maybe I should convert them all to raw MathML and remove the MathJax part. I'm a little hesitant just because of older browsers, but it does still occur to me kind of in the same way that there are a lot of diagrams in the book and I remade some of them as SVGs, but I haven't put the SVGs in yet because it's one of those sufficiently old browser that doesn't have good SVG support. A PNG is more likely to be viewable or if we turn it into a JPEG or a GIF or whatever. So, yeah, it's sort of that same impulse of, 'Ooh, I can make this be really cool, but yeah, I have to think.' But yeah, it's interesting to me that now I could, for example, drag select a section of that text and just paste it into Word or paste it into PowerPoint if I were doing a presentation and it should just work. And I want to call back the thing that you said about the LaTeX uptake, you went from zero to 30 million LaTeX pieces.
Clint Covington: We have about a million people tasting 30 million equations and we just started rolling it out.
Peter Wu: And that's the LaTeX that's embedded in HTML webpages.
Eric Meyer: Right.
Peter Wu: We believe primarily in KaTeX.
Eric Meyer: Yeah, that was interesting to me because that's already 10% of the number of PDF equations, right? The 347 million PDF equations.
Peter Wu: Yeah, no, that peaked around 500 million before the summer and it's taken a little bit of a dip, but we'll see what the numbers look like in the fall.
Eric Meyer: Right, when the students all come back. Okay, so it's maybe 5%, but that's right off the bat. I mean, most technologies would dream of even a fraction of that kind of immediate uptake. Yeah.
Peter Wu: Yeah, absolutely. And we think again, this is really changing the makeup of the content that's available on the web and in the enterprises and in the universities and such that a larger and larger percentage of it is going to have accessible math going forward.
Clint Covington: Another area that we're looking into, we don't have any announcements or anything, but I think AsciiMath is also interesting and being able to support that as an import source as well. And so, we're thinking beyond and we're continuing to look at new things we can be doing in the future.
Peter Wu: We found there's some fans of it that think it's easier than LaTeX. And so, there's a community of people that use it.
Eric Meyer: Yeah, interesting. Because I remember looking at AsciiMath when I was starting on this project and in my case pretty quickly came to the conclusion that LaTeX was probably going to be better for what I was doing or what I needed to recreate. Actually, I wasn't even doing it. I just needed to recreate it.
Clint Covington: We're working closely with Neil to bring MathCat into Narrator. So, work is ongoing there. We had it available for a few months, but we pulled it back, but we're working hard to really make Narrator a great end-to-end experience with math. And I think the support of MathCat is going to be a really critical step as we look to the future of building on the phenomenal work that Neil has done for the last decade or so about reading math out correctly.
Eric Meyer: And I'm curious, so you're ingesting these pastes and creating accessible math. What's the export story? How does this stuff export?
Peter Wu: It's a great question. So, first we focused on PDF. At least in the workflows we look at in Microsoft 365, there's a lot of people that export to PDF. It's a common target for how you're going to distribute your content. Not as much on the import side and then maybe we can do more on that in the future. We have not enabled MathML in HTML export yet, but that is something we're very interested in doing. We'd love to know if there's other apps that would consume that. So, yeah, we're very interested in learning more about that. We do have to approach it with a bit of caution to make sure that because there's a lot of apps that depend on the HTML export that we have currently and to make sure we don't break any compatibility with that. We have to make sure that we keep that kind of experience intact. So, it'd probably be something that if we ever add that we would have an option for. So, people could include a MathML in their HTML exports or not, depending on what kind of workflow and what kind of scenario they're doing.
Eric Meyer: Yeah. I mean, I was thinking of it in the context of I would like to take this accessible math and HTML that let's say Word has created or whatever, export it to HTML and put it on a website.
Peter Wu: Yeah, absolutely.
Eric Meyer: I mean, ideally there would be a more direct pipeline. I wouldn't have to paste it through something like Word, but sometimes that's how you get things done. Right.
Peter Wu: Today, you can save as HTML and Word in a way that will preserve all the fidelity when you can read it back into Word. Or if you just want to publish on the web, your main goal is to make it render well in a browser. Then there's a filtered HTML option to give you something that's a little bit more streamlined. That currently will create PNG images of your math to show in the web. But like I said, that is something we're interested in tackling next is putting MathML in there as an alternative to the PNG images or maybe in addition to it so that people have that option as well.
Brian Kardell: I just wanted to go back to the thing that you were saying about... We were saying nobody writes HTML, nobody writes MathML. You fall into a couple of basic camps. You either get some kind of WYSIWYG that has nice mapping for people who don't do programming or whatever. Or maybe if you're somebody like the people on this call, maybe you use markdown because just very quick shorthand for the simplest things that you do all the time. Maybe other people are super comfortable with LaTeX. Maybe some people are some other thing, but those are kind of like the three major camps. And then to combine that with the thing that Peter was saying earlier about we have to build all these layers of infrastructure before you can realize the full benefit and that all these pieces are starting to click together. I just wanted to mention that we have published also things to embed LaTeX in your markdown and things to embed AsciiMath in your markdown. And I worked with Murray. Murray also invented Unicode Math. So, you can embed Unicode Math in your markdown. Those are all cool and they all just generate native MathML. It's pretty nice and it should be accessible and beautiful and understandable and internationalized where that makes sense. Cool. The other thing that I can say is in things like React and View that these things that build dynamic apps, they do actually sometimes include math as well. People build these dynamic math applications. And believe it or not, they didn't handle MathML because you have to make sure that when you create these things, you put them in the MathML namespace or else they're just unknown elements. And so, my colleague Luke sent pool requests to all of those and got it. So, we're getting all of the pieces and all the layers for things that create math and they can interoperate, they can understand one another. And I don't know, I think it's just amazing. We're going to see so much more than we're seeing now. It's still so early.
Peter Wu: Absolutely. It's an amazing ecosystem that we're building here. And I love math in markdown. It's not something we support yet, but definitely something we're interested in. And that's another great way to get math into Microsoft 365. So, definitely looking at that.
Clint Covington: One of the things I've though a lot about is students in lower income countries, students from families that don't have parents that come with a strong math background. And I think it's really powerful that as we start to move forward, AI is really good at explaining math concepts. And I think all of this framework that we've been laying down provides it where students in the future who might not have access to quality math tutoring can start to use AI models to get feedback and to learn the principles of math. And so, I see AI really helping maybe close some of the gap that some students have about getting from where they are today to where they want to be from a career perspective.
Brian Kardell: Yeah. If you could believe it some of this was part of the original purposes, even considerations of MathML. Going back into the '90s, there are people who thought that AI was right around the corner. And this was spearheaded. One of the original authors was Neil Soiffer and he was with Wolfram. And they were trying to build some things that understood math, really understood it and could do interesting things by understanding math. So, that's still a really interesting bit of history, I think. But I think you're absolutely right. The benefit that we get from explaining these things in this way is that now we have ways to commonly understand them. And that means they become more explainable, they become more indexable, they become more useful basically.
Peter Wu: What an amazing journey. It's over 30 years to get MathML into the browsers in 2023 and you're hitting right at the right time. AI chatbots exploding on the scene. And now AI chatbots have math built into them and can present it in the proper form using MathML. And it's just really incredible to see. It was a bit mind-blowing. I hadn't seen it until I deep dived into math in the past year. These AI chatbots conversing very naturally with math, answering your questions and presenting it in easy to read forms. And now you can copy paste that content into Microsoft 365 apps as well. What an amazing set of things to all come together at once.
Clint Covington: I have an 11-year-old and I've been watching this journey about her moving through the different stages of math. And she pretty regularly will go into Copilot and ask questions and then sit there and read through the explanation and work to understand it. She still has to pass the test. She understands she has to learn the concepts. But she relies far less on me or her teacher to be her tutor because she does have access to this. And my daughter's probably lucky that she has access to people that can help her do that. I really think about the kids who come from inner cities and come from parents that don't have that education where I think I just get excited about the impact it has in marginalized communities.
Brian Kardell: Yeah. This I think all really, really interesting and compelling. I don't know if there are any other notes from your presentation that you wanted to make people aware of or some cool statistics that we didn't get to cover or anything.
Peter Wu: It's just really the combination of everything coming together. It was about making the math experience more accessible in Microsoft 365 apps and the output that it produces as well as making it easy for people to create that kind of accessible math. And there's a lot of different ways. And in the coming months, we're going to add more ways to make it easy for you to bring in that accessible math into Microsoft 365. And so, really one thing I want to leave with is just a call to action for everybody to check it out and to put it into practice. So, if you create any sort of content with math, take a look at how you can make that content accessible. And if you're not including math yet, maybe you should be. And maybe that's really going to take your communications and contents for the next level, really get into how things work and explain it with our universal language math. So, thank you.
Clint Covington: If I had three takeaways, the first one would be I'm just super grateful for folks like Murray and Igalia that have just been building this platform for decades that puts us in this spot where when AI comes together, we can take advantage of it. So, lots of gratitude for people that have been doing the nitty-gritty work down under the covers for decades now. The second one for me is I'm thankful to the community, folks like Daisy and Sara who've reached out to us and really shared their experience. I think that we've become better and more informed and make better product decisions because of listening to the voices of the disability community, helping us understand what's working well and where we have room to improve. And so, the voices of the community has had a big impact. And then the third for me is I'm just excited about the future. We have more things going, MathCat coming inside of Windows. I think that AI has given us tools where we can do more with math and accelerate the work we're doing in this area. And so, we have more things, maybe we come back in another show and talk more about some of the stuff that we have going on. But I'm excited about the future of math and how we're going to make it more accessible, especially towards marginalized communities and the disability community.
Brian Kardell: Yeah, me too. And I think there are a long list of things that we'll continue to do to continue to improve all these little things. Like I said, maybe this system doesn't support it. Maybe this PDF reader and operating system doesn't support it. The work is never done, but every time we do close those gaps, they kind of are pretty guaranteed to stay closed for either forever or for a very, very long time without much more investment. So, I love that we're making progress and that it's progress that has numbers that we can talk about because it's not just like, look how beautiful that math is. It's like, 'No, you can see it's in wide use. It is important. Here's what's happening.' And yeah, it's important. So, great. I'm really excited. Thanks for coming on and talking to us. I really enjoyed it.
Peter Wu: Thank you so much for having us and thank you for the amazing work Igalia has done to bring math on the web live. It's truly incredible what impact it's had.
Clint Covington: And let's keep pushing. We're not done. Let's keep going.
Eric Meyer
Brian Kardell
Clint Covington
Peter Wu