AI in Education:
A Conversation with McGraw Hill CEO

Simon Allen: We can create the most extraordinary material. But if it's not used by the teaching community and students, it's not worth doing.

Michael Krigsman: Welcome to CXOTalk, episode 838.

Michael Krigsman: I'm Michael Krigsman, your host. And we are discussing the impact of AI and technology on education with Simon Allen, the CEO of McGraw Hill.

Simon Allen: McGraw Hill has been around for just over 130 years, and we are very proud. As a company now, we call ourselves an education company, as opposed to a publisher of old that some of your viewers may be aware of, because of how we've transformed our business (we talked about that a couple of years ago).

Now, we are very much a digitally driven company that provides educational courseware materials across all of the learning lifecycle. So, we are one of the very few companies globally that goes from pre-K all the way through 12th grade into university and college level, and then into graduate programs and the medical, research, and scientific fields.

Michael Krigsman: Can you give us an overview of the impact of technology and AI today on education through your lens as the CEO of one of the foremost educational companies in the world?

Simon Allen: There has been a great deal of hysteria, frankly, around GenAI just over a year ago, particularly with ChatGPT, as it was announced a while back. And I think one has to be cautious when looking at the education environment; in the education community, you have to be very careful about how you think about new technology and advancements, which happen every day.

And you have to be sure that you are thoughtful, you're very aware of your student base and your teacher requirements, and how you think about implementing the latest technologies. GenAI is no different. It isn't about being the first, actually; it's about being the best and how you incorporate these technological tools. As we think about the education community, what we've done at McGraw Hill is focus, as we do with all new tech that appears, on where this technology can be valuable, where it actually adds expedience and real relief and joy to the teaching community, who are probably some of the most challenged and, I would say right now, stressed workers that we would see around the world. And I have the pleasure of seeing educators all the time in what I do in every country in the world.

And when I talk to teachers today, it's very, very telling just how much anxiety there is across the teaching community. What we need to do, and what we've done, is use technology in a way that reduces that anxiety, both for teachers and students, in a way that we can make sure that it is helpful to how they teach the class, and we focus very heavily on efficacy; and we need to make sure that our efficacy studies show that when we incorporate AI tools into our content and into our platform, we do it in such a way that it eases the administrative burdens of teachers.

That's a really key starting point. How do we make their lives easier, and then how do we engage students in a more robust way, which is a critical part of education? You have to keep students engaged in every subject at every age.

Michael Krigsman: You mentioned that there's a lot of anxiety. Can you elaborate on that? In a way, I understand that, of course. But at the same time, I feel a little bit of surprise that there's this trepidation rather than a sense of great opportunity and welcoming.

Simon Allen: There is enthusiasm and excitement for GenAI; no question. And we've done a recent survey, an education survey, a recent survey of over 1000 teachers in 19 different countries, which indicates that about 54% are actually very excited and enthusiastic about GenAI and how they will utilize that in the classroom.

So, the small majority are really quite keen on integrating the AI tools as quickly as possible. I think what we have to remember is that since COVID, Michael, there has been more anxiety expressed by teachers because of the additional requirements, frankly, that many of them have to cope with. And if I think about particularly mental health issues with the students we're seeing around the world, that teachers are having to deal with issues that they, in many cases, were never trained for, they were never expecting.

And the shift in student behavior has been significant recently, and it's accelerated since COVID. And I think because of that, there is a general feeling of "make sure you provide materials that help me," rather than "give me yet more to do, and give me yet more work to try and understand; provide tools that will be beneficial to the way I teach and that will add value to my students."

Michael Krigsman: As you think about the products and offerings that McGraw Hill offers, is that the kind of guiding light that you're following? Or to ask another way, what's the underpinning of your strategy?

Simon Allen: It is a guiding light, the guiding light for anything that we create. And we invest hundreds of millions of dollars every year in our platform technology. The guiding light for us is that it must be usable and useful to the teaching community and the students, full stop. If we create something, we can create the most extraordinary material.

But if it's not used by the teaching community and students, it's not worth doing.

We spend a lot of time really deeply understanding the teaching methodologies. We really spend a lot of time understanding how students relate to the teachers, and that's a key part. And as we look at our own investments across AI, and we've been doing that for over two decades, when I think about the ability to personalize the learning process, Michael, that's the key phrase, because when we launch, we have a product called Alex: McGraw Hill.

Alex, which really came out about 24 years ago. And when I look at how that personalized the learning process for math and chemistry, it was a huge step forward. And to this day, it's been an enormous contributor to developing the ability of students in those subjects where they get quite nervous, and maths is one where they certainly do.

That's using AI to personalize the learning journey. Now, with GenAI, you've got the opportunity to extend that much, much further because rather than simply thinking of STEM disciplines where you've got a right or wrong answer, and you can predict a student's performance because of the data that we have, then direct them to a particular path, which will increase the possibility of success.

And we've got lovely efficacy studies to prove that; it's great to see it. Now we can start thinking about that with more of, if you like, the soft skills issues, like writing, reading material as people vocalize how they read. And I think we can look at using AI in ways that will enhance some of the tutoring opportunities and the real skill set that students can develop, again, in the very personalized way that we could never do before.

Our feeling is, are we seeing that now with what we've tested with teachers, is that they love this because it eases some of the administrative burdens that they have. It means they could spend more time with their students on issues around extending the learning process in ways that they can help the students understand where they need correction and where they need congratulations, both using AI, and that journey gives us that much more of a personalized understanding of how the student is learning.

Michael Krigsman: Simon, to what extent do you think there is a mindset or culture shift that has taken place or needs to take place among teachers when it comes to adopting these new tools with all of the various new capabilities that they offer?

Simon Allen: I think the mindset of the teaching community is ever-changing and curious. And I think it is always keen on understanding: "How can you help me do what I do every day?" And if you can show them that you can help them, then they will engage and embrace technology very quickly. We've seen that, Michael, with the study I mentioned, where there's a small majority of teachers who are very enthusiastic about GenAI. There is also a large minority that are very keen to begin using it very, very quickly.

In AI, 40% want to recognize the value that they know they can achieve by personalizing the classroom and making sure that students have the ability to be directed and improve their success and their outcomes. I, to be perfectly honest, have a huge admiration for the teaching community. 

I was one myself many decades ago, and when I look back at that time and how simple it felt then, and yet how complicated the teaching role is today, that's why I spend a substantial amount of my time every year working with our sales and product teams, with our teachers, and with students to understand what we can do to ease their pain in many cases, as they're teaching, but also to make sure we're giving them that many more tools. And let's not forget, this is not really a transformational tool by itself; it's one of many tools because you can never deviate from the critical importance of the teacher-student relationship. That personal relationship is absolutely crucial.

And if anyone thinks that you can supplant technology in any form that replaces that, there are going to be many problems for students and the teaching community going forward. You have to use these tools in a way that augments the ability of the teacher to engage with the students on a personal level. That's where teaching will improve, and that's the only way where teaching will improve.

What we do is provide tools that allow that to happen in a more robust way. That gives teachers more time.

Michael Krigsman: Subscribe to our newsletter. Subscribe to our YouTube channel so we can tell you who's coming up on these amazing CXOTalk shows.

We have a very interesting question from Twitter, and this is from Chris Peterson. He's a regular listener, and he says he wants to know if McGraw Hill, how McGraw Hill sees the balance between paper versus online content. And will digital content take over completely, and will it necessarily be interactive to keep students interested and motivated?

Simon Allen: The business that we have at McGraw Hill now, by far, is primarily digital, and we view ourselves as a digital education company. We create all of our materials, our products, from a digital perspective immediately. If I think of our higher education business, Chris, it's now fully 90% digital, provided through courseware, one of the learning management systems that each institution may have.

And that will only increase; that number is not going to come down. I don't think it will be 100%. There will always be several students and faculty who will prefer to have print material, and we are always happy to provide that when it's asked for, but certainly, the vast majority of classes today are taught through digital materials.

There are many advantages to that. If I could just expand on that, Michael, just a bit, if Chris wouldn't mind. But the biggest advantages... So when you think about your textbook, just your print textbook, there is only a certain amount of material that you can gather, and that's what's in the book. When you think about delivering all of that content through courseware, we can embed that with brilliant assessment questions.

We can direct you to certain areas when you've done those assessment questions, to see where you may need a little bit more time to study. We can embed videos, we can embed study aids, we can embed flashcards. We can do so many more things to aid the learning process. It's what we call pedagogy. And if we have the majority, the vast majority, of our students using technology courseware as they learn the course, we have that much more flexibility to improve.

The logical tools that we can see improve the performance of the students. We sometimes track, Michael, we've seen students go up by two letter grades when they've utilized our courseware and gone through the assessments that are there. So you can see why that's so popular now. And I think, as I say, that's not going to go backward.

It will extend, I think, even further. Now, let's drop down a little bit to our K-12 business because there you've obviously got a bit of a difference where, if you think of K-5, some of the elementary education, many schools absolutely require still print products, and that's great. We, the majority of that part of our business, provide in print and very often supplement it with digital tools because, again, that's how we can enhance the pedagogical benefits and provide more material for students to use, which we hope is very valuable.

As you go up the age range, there's a very good correlation; you can see the growth in digital usage. And when you get to high school, it's really, really very, very high. And when I think of our medical business coming the other end at the professional level, pretty much all of our medical and scientific research business is now delivered online, so the shift from apparel, overall, we're getting close to 70% of our business now is digital.

That's only going to increase. The investments we make are all in digital tools and our courseware and embellishing on the pedagogical tools that we can use, of which, of course, GenAI is one of them. But that is where we refer to ourselves much more as an education company because it gives us the ability to reach many, many more students in the right way that they learn today.

And Michael, the last thing I'll say to Chris is that I've been doing this for 38 years. I began selling textbooks, actually, in Dallas, Texas. Don't be confused by the accent, but that's where I began. And, not from there, but anyway, while I was doing that, it was at that time when we were just working with textbooks.

And now as I go back to those same institutions nearly four decades later, and I talk to students and faculty, and I can see the difference in the learning process and the outcomes that the students have, and just the engagement they have is so much more robust because of our abilities to use digital technology across all age ranges.

Michael Krigsman: Your digital business today is 70% versus approximately versus 30% non-digital, and you have data that then supports the learning and pedagogical advantages to students of the digital, the digital formats.

Students and teachers, Michael. What I would say is it's obviously about the student's outcome, that's key. But we are now creating materials that allow grade school teachers to look at every aspect of the knowledge that the students have claimed. And we're doing this as a product across a whole class that we offer with some of our core textbook,

if you like, materials across the K-12 spectrum where there is an opportunity for teachers to see exactly how the students have learned a particular subject, a particular area, regardless of the grade level we can create. In doing so, a longitudinal student record that tracks the student's performance and gives the teacher an immediate glance at exactly how the students are progressing through this graph.

And when you look at the circular graph of progress, it's done in a traffic-light form, and you can see that they've understood particular subjects and components of that subject to a very granular level. That gives the teacher, and then the school, and indeed, if you wish, the school district, they can all see the progress that the students are making.

And it helps the teachers keep on track. They can identify those students that perhaps need a little bit more time and attention, but they know that the students are learning and they're learning in progress, not necessarily by age range, because we don't just want to think about "are they in third grade, fourth grade? Are they reading at that level, or are they doing math at that level?"

In many cases, they may be at eighth or ninth grade level in third or fourth grade, or they may be at first grade. It's that... the point is, what subjects have they mastered, what components they really understood within the learning objects that we create? That's what we can measure, not just for the students to help them, but critically now also for the teachers and the classroom.

Michael Krigsman: This is a question from Rashid Ghufoor, who says the advantages of AI are beneficial for everyone. But there are disadvantages as well. For example, AI is not human, and in some cases, we can't clearly understand them. So, what's your thought about that aspect of AI adoption in education?

You've got to be careful not to rush into this kind of, or any kind of, technology, quite frankly. And I would say that there are clear examples, and we've all seen them, of what's referred to as hallucinations through GenAI, where incorrect answers are given. They're given in a very confident way, which is a bit worrying. It's when the student is told that a certain answer is absolutely correct, and it's done with great confidence, they tend to believe that. Very often, we are seeing images that are completely false. We've seen answers to questions that people have posed that are completely incorrect.

And you have to remember, and I'm sure Rashid knows this, but the way GenAI works is very much from a predictive standpoint. And it's a wonderful ability for the tool to mine and respond to gazillions, trillions of bits of data and words and progressive words so that they think about giving answers to questions. And that means, as Rashid says, it's not a human element.

And you have to remember that it sometimes can appear that way, which is the danger, but it can very definitely find that the predictive analysis of the answer given is going to be flawed. That's the point that we have to think about because, and why I said to you earlier, Michael, that we don't rush into applying this new technology until we are absolutely sure that it is 100% worthy and valuable because when you have a trusted brand, as we do, if I may say, over 130 years around the world, people trust the McGraw Hill brand.

And if they see a red cube on a product, either on a screen or a print material, whatever, they immediately trust what they are going to read. We do not want to abuse that trust. We have to be absolutely certain that everything we provide is accurate and correct. And when we spend a great deal of time and money ensuring that we operate in very important disciplines like medicine, for example, we cannot make mistakes in the material that we provide.

So we need to be very careful with that. So, for example, when we recently created a test generation bank, which is, for your listeners, it's very simply a and an enormous question bank of about 3,000 questions in a very complex field of medicine. And when you create that, that can take months for a company to create.

And we decided, let's try GenAI, and we used this tool to create in minutes what would take months. But then what we did was spend a substantial amount of time, making sure with our human editorial teams, making sure that the questions were correct and that they were appropriate and the level was right, and they were genuinely testing the tasks at hand.

And I think where you can see the benefit as a company with significant efficiency on time-saving, in that example, it's clear you cannot just rely on what you are then looking at; you have to evaluate and have that human element to ensure accuracy and to ensure that what you are providing your students and your teachers is correct.

And I would say Rashid's question is an excellent one because it's too easy to rush into using this kind of technology, and if you find it creates errors, then the trust that we have with our customers disappears very quickly.

Michael Krigsman: This issue of trust and the brand is so fundamentally important. And we have a question now from Jitendriya Jamadar from LinkedIn. Continuing on the theme you were just discussing, he says, "GenAI has drastically reduced the creativity of students and degraded the standards of education. How do we differentiate GenAI work from creative work?"

I think there are two parts here. Number one is, we can debate the impact of AI being positive or negative. And then number two, as you were talking earlier, Simon, how do you distinguish between what's creative writing or hallucinations from GenAI versus what's real?

Simon Allen: I think it's unfair to castigate GenAI and accuse it also of reducing or removing creativity.

I do think the question is right in that there is a risk that that can happen if students do nothing but channel in a question that they received from their teacher, and they splurge out the answer. Clearly, that's plagiarism; that's cheating by any other name anyway. But then, yes, you would argue, and Trent is correct, that that reduces and removes all creativity.

My hope is, and I'm an optimistic fellow, and I do think there is evidence to show that what it can do, and in some ways, enhance creativity, because it's giving a lot of students a start, and what often is a real challenge for students today is "where do they begin?" And I do believe, and I'm sure you have Michael too, when I play with ChatGPT and look at some of the answers it provides, what it does is actually ferment and create the creative juices. It gets them flowing in a way that is more thoughtful and can be more creative rather than less, as long as you are a reasonably serious student. I think if you are an idle student looking for the short way out, clearly, the ability to plagiarize and cheat is absolutely there.

The good news is, I think the teaching community is beginning to be very aware when this happens. And clearly, there are mechanisms they have to track that. I would say that when a good student realizes that they've got raw material from GenAI, however it's delivered to them, then they have something to work with. They have more tools that can be relevant and immediately valuable, and it can create a very useful argument.

It can be used as a question and a way to debate, "Is this the correct solution?" And I do think that it would be unfair to accuse any technology of removing creativity because creativity is inherent in the user, inherent in each of us, and that human relationship with the teacher brings that out. So I don't believe this is a significant philosophical problem across the teaching community.

With that, the ability for abuse is absolutely correct, and that's more to Trent's point. But I would say that using GenAI, and that part of our job is to help the teaching community explain how to use AI in a way that isn't fearsome, in a way that is genuinely beneficial to the student. Again, as another tool, it's not the be-all and end-all, never will be, but it is another tool, perhaps to promote creativity, to promote the controversial questions that you may want to ask. And think of it that way, in a way that is more beneficial to the learning process. But you cannot rely just on any answer that any tool gives you, including GenAI.

Michael Krigsman: So, fundamentally, as you said, GenAI is another tool that supports learning and supports that student-teacher relationship, at the end of the day.

Exactly. And that's how we describe it, that's how we think about it, that's how we invest in it to make sure it is a tool that's adding value. And it's simply, for Michael, it has to be another tool that adds value to the teacher and therefore to the student.

Michael Krigsman: We have another very interesting, thought-provoking question from Twitter. This is from Arsalan Khan, another regular listener, and Arsalan says, "Can AI be used to rethink how and what education should be provided to students? Isn't it true that most of what we learn?" And here we have a kind of second editorial aspect of the question, "Isn't it true that most of what we learn in school, for example, figuring out the area of a donut, isn't really used or only used occasionally in our work and personal lives?"

Simon Allen: This is where I make a differentiation, if I may, between education and information. And I would argue that using ChatGPT, for example, or any number of different tools that are based on GenAI, and indeed any tool that you may look at, Google, any search engine that you may employ to look at when you ask a certain question, the information provided is very valuable.

So, you may ask then about the circumference of a donut, and you're going to get a very specific answer. My favorite happens to be a raspberry, I must admit, a raspberry sugar donut, but if you are thinking of an education, Michael, this is very different. Now think about when you… I don't know what you may have studied,

if you studied journalism, if you studied literature, if you studied corporate finance, if you studied economics or psychology, it is very difficult, I would argue, impossible, to go through an entire course that is purely delivered to you on any platform, any particular tool, when you need to really understand so many different nuances around the subjects that you're studying.

Yes, we've seen examples when the AI tool will ace the law degree, or at least a law entrance exam or a medical entrance exam. But the reality is, when you are a student learning through a three-, four-, five-, six-year journey in a subject at the higher education level, you need to think about education as part of that.

It is a slow journey that must have a very different mindset than just gathering information. You need to understand the criticality and the thinking, what we refer to as critical thought, around the subjects that are understood. You need to debate and discuss with your teachers or your fellow students about literature or about history, geography, look at what's happening in the world today.

These are elements that you cannot steer away from. And we mustn't confuse an education, which is lifelong, particularly through ages two until 32, but it's really lifelong, we mustn't confuse that with the information that we can gather from a search engine or a general AI tool. There are two very different elements, and I'm delighted that… that's a good question because everyone needs to understand the process that an education company takes to further education and improve student outcomes and therefore, quite frankly, through society.

That's our biggest mission as a company, Michael. People often say to me, "What are you really about?" And I say, "We are about enriching society because society will only benefit from an educated population." Information is very different to that, and education is a long, very thought-provoking journey in pedagogy. It's important that we remember that.

Michael Krigsman: Rashid says this. He is also concerned that AI will reduce human productivity because people give commands to AI, and the AI provides back everything without any ordinary, any extraordinary effort. And if that is the case, then where is the learning that takes place?

Simon Allen: I don't believe that AI will reduce creativity. I certainly hope and believe that it will actually inspire and enhance students to think about the answers that they receive. And if you plug in a question into ChatGPT, no matter what the subject is, the fact that you've asked the question and you're looking at the answer already indicates that you're thinking about that particular subject matter.

And I think if you've gone that far, then you will almost certainly, when you read the answer, have an opinion. You either agree with it, or you take it at face value, which would be very dangerous, and I think Rashid knows the danger that's inherent in that. Ideally, it steers and inspires you to think about what else you might want to do with that question.

When the answer comes to you, is it really the only, the absolute be-all and end-all answer? Are there more things to

There's a risk of that; there's a risk of that happening, of course. But I do believe fervently, and we've seen that in students, that when they do ask the questions, they do think about using AI, the ability for them to then extend their knowledge. And if you imagine how beneficial it is for a program, a tool, to show you where your writing skills are poor…

We've already said that we've done that with math and chemistry for over two decades, very valuable. Now we can do that with writing skills, which will be very valuable to a student who's about to go into employment, somebody that needs to really understand where they can make improvements. To me, that's the art, actually, of learning and adding creativity into how they think about what they're going to answer next, because they can learn at each stage.

So, I feel a lot more optimistic about the opportunities as long as students are taught how to use the tools in the right way.

Michael Krigsman: And I don't mean to put words in your mouth, but I would assume that the situation where a student gets information from the large language, from the AI, the large language model, and pretends that it is their own, that that becomes very much akin to plagiarism.

Simon Allen: It is plagiarism; there's no question. If the student just literally transcribes what they're given, an answer through any tool that any LLM gives out, as you say, that is absolutely plagiarism. And teachers don't take too long to recognize that, Michael. So I think when students attempt that, it will be very quick before they get called out.

And at that stage, it's a very different story for the student journey. But I do think, again, it's a requirement for the teaching community to be able to talk about how best we can use this tool. Don't use it as a cheating mechanism; use it as a way to cement creativity and begin the journey by asking the question that can be very difficult to start thinking through.

Really, teachers need to make it very clear to students. And I know through our study, particularly in higher education, there is anxiety – there's that word again – but there is anxiety with the teachers that students will be way too tempted to cheat, to plagiarize, and that they'll be so tempted they won't be able to resist. That is an issue that the teachers, and particularly professors at the higher education level, are going to face very, very keenly.

Again, it really is important for students to understand how to use this as a research tool, as the beginning of the learning process, not the end.

Michael Krigsman: So, there is a training process, both for teachers as well as for students, regarding the right way to relate to these large language models, how to use that information, and also how to be critical in evaluating the results. Because there have been many examples where people take the results uncritically.

I mean, there was a lawyer who lost a case because he got the results back in the case, quoted all sorts of case law that reported all sorts of case law that simply didn't exist.

Simon Allen: Exactly. And, I mean, that's the classic hallucination, goodness me, it doesn't get worse than that. And, you know, this is where, to your point, we're helping teachers. We provide a great deal of material to the teaching community, both at the K-12 and higher education levels, about the best ways to use AI and GenAI as you are teaching your class.

And it really is about making sure the students understand their responsibilities, how to use the tool correctly. Often they use it in the classroom directly, sometimes they use it outside, and often they'll use it for their homework deep into the evenings. And that's very important for the teaching community to understand when and how students are most likely to use it.

And then it's important for us to provide the tools and direction that teachers can use to encourage students, again, to use it responsibly. And your example, Michael, if I may say, it's a perfect one because, again, the confidence of the answer that AI gave when it was quoting laws that just did not exist was extraordinary.

And that's what that example really illustrates, why students cannot rely on the tool to provide answers, to give straight to their teachers for a grade because they are going to get unstuck without any doubt. Use it as a tool to help your thinking process, use it as an opportunity to begin the conversation, and that's what we are helping teachers understand across all levels of learning right now.

Michael Krigsman: One final question from Lisbeth Shaw, who says, "How can you address the risks of using AI? Algorithmic decision-making in education."

Simon Allen: The algorithmic process in education is clearly the predictive ability to know the answers that are going to come. And if I think again when I look at Alex, our AI program that we use in math and science for two decades plus, we know that the predictive ability is so correct that it will guide the students in the way they need to know, to direct them to areas where they need to work harder.

That is going to be even more the case as AI tools develop. That's what's exciting, Michael, that brings a smile to my face because it's really then the opportunity for us to utilize GenAI tools that will provide algorithms and solutions in an incredible way, utilizing data in the right way that would direct students to what will be a very robust journey.

Again, as long as the teaching community is comfortable explaining how they use those tools, the predictability, the cost of the algorithms is truly exciting. It never takes away the human-student interaction. Please, let's remember that the teacher-student human relationship is critical, but all of these tools around them will help significantly.

Michael Krigsman: The human teacher-student relationship is the glue and the background, ultimately.

Simon Allen: Always will be. We should never, ever deviate from that, Michael.

Michael Krigsman: And with that, a huge thank you to Simon Allen, the CEO of McGraw Hill. Thank you, Simon, for coming back again to CXOTalk.

Simon Allen: It's a pleasure. It's been lovely to see you, Michael. Thank you for asking me.

Michael Krigsman: Thank you to everybody who watched, you guys are amazing, and thank you for your questions. Before you go, subscribe to our newsletter. Subscribe to our YouTube channel so we can tell you who's coming up on these amazing CXOTalk shows. Thank you so much, everybody. I hope you have a great day, and we'll see you again next time.