Thanks, Andy. And hi, everyone.

We've delivered another strong performance in the first half, with revenue excluding OPM up 6% and adjusted operating profit growing 44% to £250 million. Assessment & Qualifications and English Language Learning have had a particularly good half, with Virtual Learning, Higher Ed and Workforce all in line with what we expected.

Profits and cash at the half year point have grown significantly, underlining the achievement of our cost efficiency program. All of this means we are confident of meeting expectations for the group for the full year and the medium term guidance we gave in March.

We're also pleased to have closed the disposal of the polls business at the end of June. Earnings per share increased by £0.031 or 14% to £0.256 due to that profit growth, with tax and interest returning to more normalized levels, as well as lower issued shares because of last year's share buyback.

Our balance sheet remains strong, enabling continued investment in the business as well as increased shareholder returns, with the £300 million share buyback commencing in the third quarter and an increase to our interim dividend by 6% to £0.07. Looking at revenue by division.

Assessment & Qualifications revenue grew 7%, driven by an excellent performance in Pearson VUE, particularly in the IT and healthcare segments. Following the disposal of the polls element of OPM, our Virtual Learning underlying growth is reported excluding this part of the business.

The Virtual Learning revenue decrease of 15% is driven by the ASU contract element of that business, which will continue to impact growth in the second half of the year. Virtual schools was supported by good retention rates, offset by lower enrollments and lower district partnership renewals.

English Language Learning revenue grew 44%, with an outstanding performance in Pearson Test of English. And Workforce Skills revenue grew 9% with good growth in both vocational qualifications and workforce solutions.

And Higher Ed revenue was down 2%, reflecting a continuation of the trends observed in the fall semester of the 2022/2023 academic year, with a decline in enrollments and a loss of adoptions to non-mainstream publishers, albeit the latter improved slightly. Turning to profit.

Group adjusted operating profit grew 44% to £250 million, driven by operating leverage on revenue growth and the cost efficiency program, partially offset by inflation. Half year profit margin grew from 9% to 13%.

At a divisional level, the same group things I've described are apparent in each division in addition to increased investment in Workforce and English, alongside some negative phasing in Higher Education relating to amortization and positive phasing in Virtual Learning relating to contract closures. It's worth remembering that the seasonal nature of the English and HE businesses mean that their margins are routinely lower in the first half compared to the full year.

Operating cash flow of £79 million was driven by increased trading profit, as well as strong collections and portfolio changes. Our balance sheet remains robust, with net debt at the end of June of £900 million.

That increase from £800 million last year is largely due to dividends, tax and the 2022 share buyback more than offsetting strong operating cash flows. In short, our operating and free cash flow performance is stronger than ever.

We've continued to invest in the business in H1, both organically and inorganically. Internally, we have continued to shift resource, cost and capital between areas of the business.

You'll soon hear about the innovations we have developed in the AI space. All of these have been achieved within the existing spend plan.

We've also made strong progress on reshaping the portfolio. We completed the acquisition of PDRI in March, significantly expanding Pearson services to the US federal government and in the recruitment assessment space, whilst also adding to profits.

The disposal of our polls business is also now complete, further focusing Pearson's portfolio towards future growth opportunities. The board has continued to demonstrate its commitment to a progressive and sustainable dividend, with a 6% increase in the interim dividend to £0.07.

And whilst we also remain committed to maintaining a strong balance sheet, given our strong operating cash flows, we can do that, invest in what we need, but also commence our previously announced £300 million share buyback. Moving to the outlook for 2023.

We are confident of meeting group expectations, with low to mid-single digit revenue growth, excluding OPM and strategic review, and group profit margin growing from 12% at the full year last year to mid-teens, benefiting from the achievement of our cost efficiency program. Importantly, as we proactively reshape the group, we are now generating some 45% of the consensus full year profit expectations in the first half of the year.

This represents a fundamental shift from where the business was even last year. We're also reiterating our guidance of mid-single digit group revenue growth over the 2022 to 2025 period, and for margins to rise to the upper end of mid-teens in 2025.

In terms of divisional guidance and phasing, there are a few points I should highlight. We are raising our revenue growth expectations for 2023 in English Language Learning from high-single digit to around 20% following strong growth in the first half, which we expect to normalize in H2.

We're investing a portion of our operating leverage on this improvement to support future growth opportunities. In Workforce Skills, given the considered approach that we are taking to ensure we have the right go-to-market strategies to scale that revenue in Workforce Solutions, our 2023 and 2022 to 2025 growth guidance is likely to be more stretching.

Continued growth in Pearson+ subscriptions will lead to a shift in Higher Education revenue recognition from Q3 to Q4. Contract timing in A&Q will see delivery in earlier quarters, meaning Q4 revenue growth will be lower than average.

We're confident in achieving full year guidance. And it's also worth flagging that the termination of the ASU contract will continue to impact growth in the second half of the year for Virtual Learning.

In virtual schools, our expectations for revenue growth remain unchanged, with H2 enrollments impacted by the loss of that large school contract we've previously discussed. I appreciate there are many moving parts given the recent portfolio changes we have.

But to put things simply, there'll be no revenues or profits relating to OPM being ASU plus polls, or the strategic review businesses from the beginning of 2024. So, in summary, we've had a really strong first half performance in all financial metrics, and are confident of delivering on full year expectations.

And with that, I'll hand back to Andy.

Thanks a lot, Sally. Now, I'd like to take a deeper dive into our AI capabilities and highlight some of the things that we're working on.

We're still very early into the generative AI journey with new developments happening on an almost daily basis. So it would be wrong to claim victory so early in the game.

And we're also very aware of potential risks and turbulence ahead. But to be clear, we see generative AI as being a real long term positive to the company, and not just in Higher Education, but across our portfolio of businesses for the following reasons.

We believe the value of our proprietary IP and datasets will increase over time. We have deep AI experience and expertise across the whole company.

We're starting to introduce new AI enabled products across the business and we've already achieved [Technical Difficulty] but have yet to commit to. What this interest does demonstrate is the real value to be had of owning your own intellectual property.

We're also continuing to monitor legal and legislative developments very closely. Now one of the biggest areas of focus for the team has been around the positive implications of embedding generative AI into our higher education courseware.

We've been working with one goal in mind, namely, how to improve the learning experience for both faculty and student. As I mentioned at our recent update, one of our key advantages is the quality of the datasets we're able to use in generative AI due to the ownership of our IP.

Simply put, the purity of any data input equates to greater accuracy of any data output. As you'll shortly see, we've been working on ways to embed generative AI within the learning experience.

We think it's very important to utilize this technology in a frictionless and seamless way to benefit and enhance how students learn. Core to everything that we do is maintaining trust and quality and generative AI is no exception to that rule.

We're not interested in utilizing this technology merely to provide students with a shortcut to an answer. When we tested different LLMs with a question from Campbell's Biology, they often didn't get it correct.

So we believe delivering product that is reliable, accurate and trustworthy is paramount. Now, here to give you a quick insight into just some of the product innovation we're launching for back to school, I'd like to hand over to our new Chief Product Officer, Tony Prentice.

Thanks, Andy. We're now working on adding generative AI to our products across the company.

But for now, I'm here today to show you the AI features we've been working on in two of our key higher education products, Pearson+ and MyLab & Mastering, both of which will be available for this year's back to school. The goal of these AI study tools is to help students learn and actually understand tough subjects.

We want students to stay and learn inside the Pearson product environment. Let's jump in and start with Pearson+.

Let's imagine I'm a biology student, and I'm using Campbell Biology and Pearson+. Just want to note that everything you're about to see is being offered in beta in a small number of titles for back to school.

My professor has been teaching genetics. I want more information on what I heard in class today on recessively inherited disorders.

I see the icon in the bottom right, and I can click there to access the AI study tools. I ask it to summarize, and immediately it does for recessively inherited disorders.

Okay. Next, I want to learn more about dominant disorders.

So scroll down to that section. And I realize I learn best when I can compare and contrast.

So I ask the AI to do that between recessive [Technical Difficulty] little more practice on this topic. So I can ask for practice.

And it gives me questions as well as answers, which is one of the things students ask for the most. So it was really cool for me as a product person that we're taking the Campbell Biology content created and refined over the years by subject matter experts and combining it with the conversational nature of large language models to break down the topic in a really personalized way to help students actually learn.

Maybe I need some more help to really understand these concepts. So I stay in Pearson+, but move over to channels, where I see the corresponding section from Campbell's Biology.

I can see this new feature on a tight schedule. It uses AI to summarize what's in the videos, so I can see if it's the information I need.

Yep, this looks like it. So I'm going to go in and watch the video.

And after that, I can scroll down and get yet another set of practice questions. So that's Pearson+ and our new embedded AI tools.

So now we're in MyLab & Mastering. Imagine I'm a chemistry student, and I'm trying to figure out this problem, how much mass is converted to energy.

As you can see, this is a pretty complicated problem. I think we covered this.

So let me give this a try. This is my first try.

And I can see I got it wrong. You might remember what it was like to study science, get like a really complicated problem wrong, and how frustrating it was not knowing how and where in the equation you were making the error.

But now I can ask the chatbot for help. So I click on the button for help.

And what you see is a series of smaller bite sized questions. We're not just giving the student the answer.

We're helping the student work through the steps to get to a solution. All right, so let's see.

So answer the first question, and then you can see another problem that goes a little deeper. We're going to ask questions to gently nudge the student through the problem.

This is what we're trying to do to help the student actually do the work. And again, this continues for a few more steps.

Eventually I'll have all the concepts that I need to try and do the final answer again. So now, I'm going to go back and put in my final answer.

Let's see. Sweet, I got it right.

Scientific subjects like this are best learned by completing practice problems, solving equations and working formulas. Students often get stuck in these complex problems and don't always have the right tools to really work through them.

So instead of students having to wait for office hours or just giving them the answers, we're introducing this AI assisted learning tool to help them learn at their speed. What you just saw is just the beginning of our AI product journey.

This is version one. Generative AI is rapidly changing and advancing.

And as our own testing and research evolves, you will see our products and features evolve over time to meet the needs of students. I look forward to showing you more as we continue to learn and innovate.

Thanks, Tony. And by the magic of television, you transported yourself from Hoboken to 80 Strand, so Tony's here for Q&A, both for you online and for you in the audience.

And welcome again. And thanks.

More from Tony later. Okay.

As you can see, this is just the start of what I believe will herald an exciting future opportunity for Pearson as we thoughtfully utilize the capabilities of generative AI across the whole company. As you can see from the examples in Pearson+ and in Mastering, the ability to give students a highly personalized experience is the cornerstone of the AI product, helping students learn in ways that best fit their unique needs.

It's just one of the ways we can meet consumers' expectations and build a further trusted relationship with them. So, in conclusion, I couldn't be happier with our position today.

And I'm optimistic about our growth prospects going forward. As you've heard throughout today's presentation, we're executing well on our strategy, and our financials bear that out.

We have a stable, growing and more evenly distributed portfolio of businesses with learning and certification at the heart. We see a growing number of opportunities to leverage our IP and datasets to create compelling and innovative products, incorporating generative AI.

And all of this means we're well positioned for the emerging future of learning. With that, Sally, Tony, and myself will be happy to take your questions.

Jo, over to you.

[Operator Instructions]. We'll go first question from Matt Walker.

I've got three questions, please. The first one is on Higher Ed.

There was a bit of an improvement in Q2. Maybe you can explain what was behind that and how does it impact the full year?

Second question is you're starting the buyback in Q3. Do you have a rough idea of when you're going to start it and how long it's going to take to complete?

And then on A&Q, you had a good result in A&Q. Is the 7% roughly a good figure for the year?

Or is there any reason they will slow down in the second half?

Yes, Q2, we saw growth. You'll remember that we spoke at the beginning of the year about Pearson+.

And as we see growth in Pearson+, we see those revenues come in as the subscription comes in. And so, we're always expecting some of the revenue from Q1 to be in Q2.

That's what we saw when we looked at Q1 results and you see that coming through in Q2 as well. And we're confident of hitting that full year guidance with an improvement in Higher Ed.

For the buyback, obviously, we announced that at Q1 trading. I've waited to start it because I'm drawn on my RCF at the moment.

That's just the normal kind of pattern of cash flows for Pearson during the year when given that the we did the share buyback last year as well. Half one cash has been really excellent.

I'm really pleased with that performance. And that means that rather than [indiscernible] H2 at that point, now we're talking about Q3.

So it's imminent. Paperwork is on my desk.

We'll be ready to go soon. A&Q, really pleased with that growth in the first part of the year, particularly the growth in VUE that's come through there.

I have pointed out in the kind of detail of the guidance that q4 for A&Q will be a little bit lower and that's just purely because of where some of the contracts have fallen and rev rec over that, but we're really confident about full-year guidance frankly as well.

Very briefly on the Higher Ed piece, as mentioned earlier, I take note of the uptick in retention rates into the 90% plus range from the 80% plus range. That shows that more faculty.

We have great product market fit [indiscernible] often a leading indicator that shows that before we've done generative AI improvements and enhancements, which we think are unique in the market, we've had excellent feedback from faculty. We've been testing this with both faculty and students over the past several months.

I think that's a really good indication. And then Tom and the team, as I referenced as well, we've reorganized the sales team and we're seeing real impact from that as well.

So as you know quite well, it's a process as we're going forward, but some of those indicators, I think, are quite positive.

Can we take the next question from Nick.

It's Nick Dempsey from Barclays. I've got three left, please.

So first of all, can you maybe explain in a bit more detail the exact change of strategy we're talking about in Workforce Skills? So in other words, what you were hoping to achieve with the skills platform and what you're currently now hoping to achieve and why that is a change to your hopes for revenue growth?

That's the first question. Second one on cash flow.

It was pleasing to see in the first half that cash product development spending was pretty much in line with the amortization in the P&L. Is there some timing there?

So should we assume a notable gap between the two, like we had last year which was £54 million? Or is this the new normal?

And the third question. So in the polls business, I think, Sally, you mentioned a positive effect from contract closures, giving you a bit of a positive bump up in the first half.

Can you just explain what drove that and how it affected revenue or profit?

Should I take the first one and you can take the other two, Sally. In short, there isn't a change in strategy in Workforce Skills.

What we're pointing out is there is a change in the demands of customers in terms of how they want to access the portfolio of workforce solutions that we're providing. And I wanted to get away from – in essence, when we showed the concept car video at prelims that people thinking of this as a one set product, you shouldn't think of it as one set product.

You should think of it as a set of customizable solutions and that vary and are applicable on a company by company basis. One of the things that we're finding is more and more companies become more and more engaged in the upskilling and reskilling process.

They themselves are learning a lot more about what capabilities they need, and the solutions that they need vary company by company. So rather than have the impression that this is a one size fits all, I think as I referenced in my script, you should really look at a suite of the talent investment platform and – TalentLens, Faethm, Credly as a suite of solutions that we're able to offer.

And as I say, we had 12% uptick in the number of enterprise clients, including many leading Fortune 100 companies. So we're getting scale, we're getting feedback, it's about meeting the customer where they need to be, and being agile and fluid.

And I think that is a theme that you're going to see happen more broadly as we move from this role. Even take higher education, a textbook was a fixed product.

And now we move into the digital world that you see of constant iteration and change. So that's the sentiment behind what we mean in terms of the shift to talent investment platform.

And it allows us then to really flex and focus on delivering the needs of enterprise customers, whatever they happen to be in each of their own unique circumstances.

On the cash flow, well spotted in terms of the first half cash flow, and yes, I think we have talked about that theme of product development, amortization and spend converging more over kind of future periods and actually for CapEx, that CapEx number coming down compared to where it had been previously. And then, for polls and that contract piece, effectively, we've got from a revenue recognition which drops through 100% to profit, which is why I'm pointing out from a profit point of view, more from a comp basis in the first part of the year, that's just to do with how the revenue recognition worked on a couple of contracts that we closed out before the disposal part of the deal.

And then you'll see that, from a comp basis, reversing in the back part of the year. And then we've got exactly the opposite happening from that Higher Ed phasing piece, a negative impact in the first half of the year, which, again, will reverse out in the back part of the year.

So it's just in the detail when you look at divisions.

And we'll go to Adam in the front.

This is Adam Berlin from UBS. I've got three questions.

First one for Sally. Can you confirm that the H1 performance was above budget?

And if so, why are you raising guidance for the full year? What are you worried about in H2?

My second question is about Higher Ed. I think a few of you could answer this.

How important do you think it is in this market to have the best product rather than the cheapest product? What gives you confidence that all this investment you're making in product development is going to work for adoption share?

And do you expect to see adoption share gains this fall? And thirdly, a question for Sally, can you give us free cash flow guidance for 2023 please?

You can take the middle one and I'll take the first one.

Yeah, I have a comment on the first one, but you kick off.

I won't get into our kind of internal budget. But, clearly, we're really, really pleased with the first half performance, and it's better than maybe people would have expected, which is great to start off the year with a real positive.

We're only half one. But I'm more confident than ever of hitting that full year number.

And then, in terms of free cash flow, and then I'll let you come back to the middle one, so we've guided from an operating cash flow perspective to 90% plus. Then from a cash perspective and an interest perspective, you'd see similar numbers to what you would have seen before.

And then, the only other thing to think about is those reorganization costs that we had at the latter part of last year. I'd guided to £85 million, I think it is, of that being in cash terms.

The rest was just P&L terms. I think we've seen about £40 million of that go through it in the first half of the year, the rest will go through in the back half of this year, and maybe a little bit into next year.

Quickly back to the guidance question first, Adam. We probably – when you have time, I realize it's early, study the numbers, English Language Learnings, for example, we had – and Sally referenced it, in first half, we took some money out of the increase in revenue and reinvested that into providing future growth, both on the ground and preparation for what Canada [indiscernible].

So we could have delivered higher than the £250 million in H1 if all we've done is focus on a number and hit a number without having to think about also being more flexible and fluid about how we manage the company for consistent sustained future growth. And you're going to see that pattern continue.

That's why we're reiterating how comfortable we are with guidance for H2. But rather than stick a number out there, I'm sure all of you who are far smarter than I will come to your own conclusions.

I want to be in a position where we're building this company to deliver sustained growth and remain flexible and agile to react to whatever's happening in the market across our whole portfolio of businesses, rather than just being focused on hitting a number. Because that's really short term thinking.

And we are in the business of really trying to grow sustainable growth. That's why we gave and said what we said.

You shouldn't read any more or less into that. Certainly, we see, all things being equal – as we sit here today, we don't see any headwinds or banana skins or whatever in the second half of the year.

But we want to remain thoughtful, as we always have, about how we manage this company and how we continue to grow – deliver growth, which we have demonstrated I think quite a good track record over the last two or three years.

Actually, I'm going to pick that up as well because you asked about risk and I didn't particularly address that. There's no risk that I'm worried about in the second half of the year.

In fact, I'm more confident than I would have been in previous years. So…

I guess. Don't overthink it in a way.

Quality always wins. Doesn't matter what you're in.

Quality always wins out. And quality doesn't necessarily mean the most expensive either.

You look at the Pearson+ and the threefold increase in subscriptions, where actually we're offering really high quality and additional features in addition to the e-text within the Pearson+ environment, all at a very affordable price. And so, that's one thing, I think, that we really focus on trust and quality.

The second thing is, what we do is really hard. And if you're staying for the demonstration after this, when we're going to get a bit more with Tony and I into our future roadmap and generative AI, we have a great responsibility on delivering a word that I hadn't really come across before I joined the company – efficacy.

It's okay having something cheap and not very good if it doesn't work, particularly when it comes to learning and particularly education any form of learning, but particularly in education. That's not good for anyone.

And it's not a long term solution. I think what you're seeing actually is, in Higher Ed, some of the rebound from we think lower adoption of faculty of OER and what I called, I think, in one of the previous calls, homebrew where, during the pandemic, faculty were kind of making their own stuff up, a bit of this, a bit of that, bit of the other.

Guess what? They found it didn't work.

It wasn't as effective. Didn't provide the learning that faculty wanted to deliver to their students.

And that's why we're going to remain focused, first and foremost, on quality and trust, and secondarily, on price because the two are not necessarily – equate to each other. So makes sense.

Can I add just one thing? I'd just add one thing on that.

We've been talking a lot to students, and the number one ask they have is to help get better grades in my class. And so, that's what they're asking for.

And so, I think, again – and we're going to get into this, it's quite hard, but focusing on how do we help them actually learn. And they know that it's actually quite tricky.

If you saw in the video, especially with some of the scientific questions, when you're in the middle of it, you get stuck. It's just really hard and really frustrating.

And so that, to me, is what they're telling us where they need help.

Let me go to Tom Singlehurst.

Tom from Citi. A couple of questions.

Maybe three. Let's see.

On English Language Canada, it's clear that we shouldn't factor anything into our numbers for this year, but can you just give us a sense of, A, the size of that market and then, B, how market share may be expected to ramp based on, let's say, the Australian precedent. I don't know what the share split is between you and ETS and Cambridge in Australia.

That would be interesting to know. Second thing, on Higher Ed, it sounds like you've done your bit on to upstream, making sure you've got the adoption in place where you want it.

Have you any thoughts at all on how many students are going to turn up in September? That'll be interesting.

And then finally, for Tony, on the AI investment, they look like very cool tools. Are there going to be included in Pearson+ as well as MyLab & Mastering?

And then how long will it take you to ramp it up to the to the full suite of content.

First two and obviously the last one is for you. We're delighted to have got recognition in Canada.

And you'll notice it came in two pieces. So the immigration part first, and then the really important student piece second.

There isn't any market share data. So we have to estimate it, but we think it's around £200 million to £250 million, possibly a little bit higher.

And these things do ramp over time, but it's going to be a reason that we'll see good growth in this business and then PTE specifically, as we go into next year, and quite a few beyond. You look at Australia in the period that that ramped over, that feels to me like it's probably a good way to estimate that.

Higher Ed enrollment, so, yeah, we're pleased with, to your words, the things that we've done in the first part of the year. So that performance and that sales team that we restructured at the beginning of the year and seeing some proof points in terms of data coming through.

We're pleased with what we've done on product from a stability point of view through some of the new things that we're putting in that were planned this time last year and some of the stuff that we're putting in from an AI point of view. So lots of reasons to feel good about that.

From an enrollments point of view, you all have seen spring normalized a little bit. One thing I've learned is not to call what enrollments are going to be for fall/back to school.

We'll find out in September/October, I think, is the way to think about that. And the AI one's for you.

As we mentioned in the video, what's really interesting is we have this amazing IP with this amazing expertise. And then we have some of these great conversational nature of the LLMs.

And so, what we're doing is trying to take advantage of both. What you saw, we're launching, on Pearson+, three titles.

I believe it's Tro Chemistry, Biology, and Kotler's Marketing. And the plan is to really put it in its beta because it's not all the way there yet.

And so, what we're going to do is we're going to put it in and we're going to learn with students, and I think being on the field, rather than being on the sidelines is the exact right place to be right now. And so, we're only going to do a limited number of titles, and we're going to be all over maniacally staying close to the students and then we're going to iterate.

The models are going to iterate and we're going to iterate. And again, we're in the business of helping students learn.

And so, I think what you've seen is us jump starting already ahead with a great solution. But staying really close, we're going to agile, kind of like learn together as we go through this.

And we'll take the next question from James.

It's James Tate from Goldman Sachs. Just one question, please.

Could you give some more color on the benefit you saw from the cost savings in H1 compared to the £120 million you've talked about for the full year, just trying to understand sort of the phasing between H1 and H2 there?

So, £120 million for the full year. So, half of that £60 million, we were almost at £60 million.

So high 50s in the first part of the year. So all we have to do in the second half of the year is just execute the same things in the second half of the year.

So, those cost savings, super secure.

We'll take the next question from Luke.

This is Luke Holbrook from Morgan Stanley. Just a question really on the licensing, you said you've been approached by multiple parties.

What's preventing you kind of go down that route? And can you elaborate on some of the conversations that you've had in that regard?

And just secondly, it's some interesting, I guess, product launches that you're doing on the AI side? Are you looking to try and commercialize it further through product pricing?

Or is this just being wrapped into the existing products' fee structure that you have?

I can take the first one and do the second one. I don't want to just take the first offer that comes along.

And because, as I said in my script, I think the value of these datasets actually increase. And one should think about the datasets we have around Faethm, the datasets we have around Credly, everyone thinks about the datasets that's just within our higher res [ph] courseware.

Because what we have is really valuable. It's easier, I think, for us to comprehend with a book, so I use that as an example.

These large language models are being trained on ingesting words and facts and things like that. Now, if one were to take a light novel, nothing wrong with Mills & Boon, but a Mills & Boon type novel and ingest that, that is going to deliver a certain amount of output.

You get a lot more than words with our product because as you even saw in that one example, the complexity around the construction of how we develop those textbooks, which has been developed over decades is very, very valuable. So we're not in the business of just selling content for the sake of selling content.

We want to be very thoughtful and specific as to what we get out of this versus what a third party LLM gets out of it. There are some interesting opportunities we have that present more of a walled garden approach.

There's some broader opportunities, but the space itself is moving at a highly, highly fast pace. And so, being first and just announcing a deal for the sake of announcing a deal is probably, in hindsight, you may look back and go, that probably wasn't the smartest thing to have done at the time.

And so, it's not for a lack of interest and it's not for a lack of thought on our side. It's just that we want to be prudent in terms of – because once you put this stuff in, it's hard to get out, as you know, and then there's – this isn't dictating us, but it is still early days in the litigation and legislation areas as well.

And that's moving at a pace. So being first is not necessarily the best long term strategy with this, utilizing what we have in products like what you've seen and constantly iterating is the way we think to go forward.

But it opens up a lot of – to reiterate myself, we had a lot of really, really valuable datasets. And if you want to add, Tony, or…

I've been here for about three months. And one of the things I've been really pleasantly surprised by is, first off, how much existing machine learning, data science, AI already existed at the company.

I think you saw that in the videos already starting with Pearson Test of English. So I think there's a lot that we always had and have in our DNA.

Then as we look forward, I think our goal is to focus on helping students learn. And so, what we'll do is we'll look at how we can bring these together.

But don't forget, there's commercial LLMs, there's building our own, there's some of these smaller models. So there's a whole space here that's evolving.

And so, what we'll do is we'll continue to be evolving, and we'll make sure our customers have the best of both worlds. And then we're always really looking at value propositions.

And so, we'll always be looking at how do we solve those problems and what's the right value prop that we bring at it. So those will all be what we're looking at.

And again, that's why on the AI thing, specifically, we're only going for a few titles, so we can test and learn and be really agile as we go through this.

And on the pricing question, of course, very simply put, yes, we could extend that through this. But there is actually pricing increase in Pearson+ at this back to school.

And for those who are not taking LLM, there's a pricing plan going to be introduced for channels as well. We've seen really good traction for channels in over 170 countries around the world.

And you saw how we're integrating generative AI into that feature. So as Tony says, let's learn, let's iterate and clearly we then have pricing opportunities.

And as well as pricing, of course, those added feature and the extra value there is for the student will also start eating into the secondary market and people come into the product because you're getting something that you can't get. You just can't get it from a textbook.

When you saw it on the screen, it's super cool. You can't get that from a textbook.

Yeah, that notion of not leaving your environment is really important. You don't have to click out to click into somewhere else.

It's all embedded. It's all within the one environment.

That's really, really important.

We've got a question from Ben.

Two questions, please. Firstly, just to clarify on the English Language guidance, you previously guided to high single digit growth with improved margins.

And now you're highlighting that you're investing a portion of operating leverage. So should we be assuming a similar level of operating profits despite the high growth?

And then secondly, on AI, could you describe how AI additions place Pearson's offering against peers? And I'd be really interested to hear how you think Pearson's offering stacks up against peers today and what the vision would be over the next couple of years?

The guidance effectively from a profit perspective would be the same. So we would have had high single digits from a sales point of view now.

It'll be circa 20%. You'll still see an improvement in margins.

And then that investment, we've just taken some of that sales extra. So the difference between the low single digits and 20%.

And rather than seeing that drop all the way through to sales, we've invested some of that in two things. Marketing, we're seeing markets like China open up.

We want to make the most of that. And also, in some places, we're getting to capacity, nice problem to have.

So the kind of VUE test center network, opening some new test centers across India to make the most of the extra volume that we've got, particularly as we start to think about Canada coming online as well. It'd be a real shame to hit up against capacity.

We want to make the most of all the great work that we're doing.

It really plays into Adam's question earlier around the broader guidance. Investing now for the future to maintain sustained growth, I think, is really important for this company.

Was there a name of a specific competitor that you were saying? I missed the last part.

Just in general, okay. I think if you think loosely two big areas, there's the over-the-top commercial LLMs, Bard, ChatGPT, kind of things like that in one space.

And then you have closer in edtech. I think from the broader stuff, what we've seen is more, let's say, great, amazing LLMs that can answer questions, but not necessarily targeted towards helping students learn, which is where we're focused.

So that's kind of how we see that. I don't think we see anybody in the edtech space yet, but I think we have to assume that there'll be stuff coming.

So I think what you're seeing is actually quite differentiated for now. And I think the whole point of us being in market is to try to stay ahead.

So that's how I would think about the two sides of the equation. And I think that's how we're going at it and making sure we're there now.

As we said in the prepared remarks, we're not in the business of just giving students answers because that's very short term. And actually, the research – students realize that.

The more and more they're going to realize that actually it's the learning piece of this that's critical and very important, and that's why we're focusing very much on that. 4,009 students, I believe, is the exact number that we've talked to and been iterating with.

Why wasn't 4,010 is beyond me. But getting feedback and getting constant iteration is really important.

And for those of you who are able to stay after this presentation in person, we're going to get a bit into the future. We're starting in higher ed, just as we did with Pearson+ in higher ed because we think it's interest, but you can imagine us being able to, and you saw some examples – if you think about Pearson as being in the learning and certification space, delivering a certificate every 1.6 seconds and having this mechanism, whether from Pearson Test of English, through Pearson VUE to Credly and the content that we're able to create, applying generative AI across that suite horizontally is very, very interesting for us as a company, and that's unique.

No one – we have no peers who can do that end-to-end solution. That is unique.

That set of assets is quite unique to us. And you then start to utilize generative AI on an individual, as you as a learner to help improve you as you go through your career.

We will get more into this after because there's some stuff that we'll say offline. Any questions, Jo, from online?

Yes, another two. So, let's start with one of them.

Analyst has asked how should we think of slower growth in Workforce? You can answer that question, then I'll come on to the next one.

I don't think we said slower growth. I think we've said that – and I'm not changing the guidance that's out there, which is double digit this year and more than 20% over the medium term period.

I'm just pointing out that as we work through making sure we're really delivering to customers what excites them that it's a little bit more stretching. So I didn't say slow growth there.

And three questions from AlphaValue. So the first is, are the generative AI tools developed on open source LLMs?

Second question, do you run them on your own servers? And the third question is, do you have your own team for machine learning?

I'm going to take it backwards. We do have our own team.

Although to jump to the first question, we're using – for the ones that you saw both on MyLab and on Pearson+, we're actually using ChatGPT behind the scenes. We're using it through Azure.

No, we're not running it locally on our servers, although we are also testing almost every single open source model out there against the commercial ones. And what we'll continue to do is look at performance versus cost as this all evolves.

But, specifically, we're using the commercial ones. I don't know if this is what was behind the question, but we take data privacy, data security – incredibly, incredibly important.

And so, we're using Azure in part to help us. And one of the most important things that both Azure and GPT will do with us is when students put their data in that won't be used to train future models.

And that's an absolute criteria for us as we're going through this. I don't know if that was hidden underneath the question, but that's an absolute kind of principle for us

And with that, Andy, I'll hand back to you

Well, thank you all for those of you who have joined us online. Appreciate your interest in the company.

And thanks again for those of you who've joined us in person and we will look forward to sharing more about gen AI with some of you here very, very shortly. And see you in October.