Watch the video:
Read the transcript:
Since it figures into everything that's going on in education, what’s your take on AI and education?
Chris O'Malley: Artificial intelligence and education–it's been a little bit controversial and I think it's been portrayed one way, which is why it's been that way. And I'll agree with many pundits that if a student is using AI to do their homework for them, that's not good. It's not good for their critical thinking, et cetera.
But what I'm focused on is using AI to make the teacher a better teacher. I think it is probably the only way we can personalize learning for all students. A teacher simply does not have the time. If they're a teacher with 30 students in class, they have to do all the prep work, all the homework, and then in the teaching itself, which is the fundamental part of what they do, they cannot personalize it all on their own.
If you can take advantage of the goodness of AI, you can now personalize the learning. And we know very, very well that personalized learning is a dramatic difference maker for education. There's a famous study, the two sigma study where if you're a teacher and you have 30 students, you typically are going to have a standard bell curve of your students. There's the mean and then there's the people on the far right tail and the far left tail. We know that if every one of those 30 students had a personal tutor, the standard deviation is going to move two sigmas to the right. So the mean of the students is going to be where the far right, your best students were in the one to 30th configuration. So we've known this for years, it's called the two-sigma problem, but it's just impossible to personalize that. We cannot afford tutors for everybody to do that.
With AI, it's the first time we have the opportunity to personalize learning. I'll give you one quick example for this.
Say I give you a math test and you get questions eight and nine wrong, and it was a ten-question test. Historically, the teacher would say, "You know what? You got an 80%, you got to improve." Something like that.
If AI knows the context of every one of those questions, it would say, "You know what, you understood algebra, but you just didn't quite understand the associative principle in math. Those two questions that you missed, 100% related to associate principle.” The AI can then provide the teacher with new concept materials for associative learning and 20 follow-up questions to make sure the student becomes an expert in associative learning.
And that's the difference where you personalize that learning. You're going to really see the improvement in education.
It fits in with that push toward mastery learning rather than just grades, and deeper understanding of a subject that leads to better outcomes as well.
O'Malley: So the two sigma tests that I referenced, the standard one to 30, mastery learning basically moves you one sigma to the right. Personalized tutoring moves you two sigmas to the right.
Most of your good teachers are going to follow mastery learning and they're going to move that one sigma to the right. But with AI and personalized learning, it can move two sigmas even better. And before the advent of AI, that was really impossible to do.
I think now over the next five years, educators, we're going to figure out how to use AI for the goodness of the students, the goodness of the teachers, school districts, et cetera. It's great potential for that.
So in that vein, what are Google's class tools and how does running them natively on a Chrome OPS module change the teaching experience?
O'Malley: Obviously, Google's got a wonderful infrastructure–you've got Google education, there's a whole unit of materials, content, teaching built around the Google ecosystem. Historically, they've had a Chrome box that helps with the teacher. They've essentially taken that Chrome box and put it behind a flat panel display and that compute now makes the flat panel display an integral part of the teaching.
The way things used to be, maybe you put a slide up there. It's not very interactive, it's kind of interesting, it's difficult to make out the content. But when you throw a compute on the back of a flat panel display, all of a sudden you've got something where the content can dynamically change. And, whether we like it or not, our students have learned to consume information through media.
To bring them into a classroom and you have zero media, that might be a challenge. I don't know that you want all 30 people with a device, that could be very distracting. But if you have one device with the computer on it that's controlled by the teacher with educational appropriate contact and the ability to interact with it, that's really good because students can consume the media in the way they've used to–rich graphics, videos, gamification, all of that can be done.
And then when you have the goodness with the Google ecosystem on it, all of that stuff that they have with Chromebooks, it interacts very nicely with the Chrome OPS. You can have Quizlets, everything like that. If I’m a student holding my laptop here, the laptop and the IFPD or the flat panel display can communicate.
So it's a great infrastructure or ecosystem to add to the classroom. Then what's really nice is you still have that compute there. So there's a lot of AI workloads. People think you have to do AI in the cloud, you don't. You can do AI very locally and the Chrome OPS is a powerful system and you can access AI right on that.
Can you provide an example of how that might go?
O'Malley: So I think you can dynamically change the content of what you're presenting to pull it in. The flat panel display with the compute is going to have access to the entire Google ecosystem and all of the material, the training and education material that the teacher wants, and it can dynamically pull that while you're presenting.
The key thing we found with the flat panel display is it's got to be part of your pedagogy. It's got to be integral to how you teach. When it becomes the critical thing of how you teach, it's really, really valuable. If it's incidental and you rarely use it, it's not great. But if you make all of your content materials interactive with it, it really brings out the interest of the students and you see significant improvements in engagement.
If you have a static poster board, people's attention isn't drawn to it as much. But if you have a video screen that's constantly moving and dynamic, people actually engage with it. They're going to look at it longer. We can take advantage of that in education, not to distract, but to teach and to educate. And that's a powerful tool.
From a security perspective, what are the advantages of managing a panel via the Google admin console versus maybe a separate standalone operating system?
O'Malley: With AI, most people are thinking ChatGPT or Gemini from Google where it's 100% cloud and that's great. For a lot of our work, it works really, really well. There's reasons you want to go to the cloud. You have access to an immense amount of compute potential, but you do have security challenges. So student privacy is an issue. You may not want student grades, student names, student images to ever make their way to the cloud.
So the Edge is becoming really, really important for that. Let's say a school district or a school itself could have certain security protocols where they say, "I want no student grades, no student information, nothing like that to ever leave the premises of the school." And then you can write a hierarchy, say, "It's okay for some data to reach the cloud, but I'm going to limit what can reach the cloud so there's no student identification information reaching the cloud." So all of that with Edge compute becomes very important.
If you have a local Chrome OPS behind a flat panel display, you can reside it there, but you can also have Chrome boxes or the teacher's Chromebook sharing that information constantly, but you can keep it on the Edge where you know you can secure it. That's something we at Intel are working on as well. We'll work with people such as Google where we'll create what we call a hybrid AI structure where what should be on the Edge has latency or regulatory or privacy requirements. But certain things that are better done on the cloud, let's enable it to go to the cloud provided the proper security safeguards are taken.
We're helping the industry build those safeguards–when should it be done on the Edge, when should it be done on maybe a server that sits in the school district, and when should it be on a cloud that's a public cloud that maybe has more security challenges with it. Those are the types of things that we're working on with the industry.
One thing that is important as well, you don't see it so much in the U.S. but a lot of countries now are starting to actually regulate access to the AI, the clouds, the chatbots. So a lot of countries have, if you're under 13 years old, you can't access that. So if you’re a school district, you can't really rely entirely on the cloud for that, you've got to have Edge compute to do that. And then the goodness of that, and if you follow the AI industry, is that there's an immense amount of power that you can now do to free local models, small local ones that you can run AI in the Edge.
That's what's going to enable a lot of these districts who have these privacy rules with ages 13 and below. You can keep it entirely local to do, and that's a big, big help I think for the entire educational industry.
How does a certified update lifecycle for a Chrome OPS module help districts plan for the future?
O'Malley: A Chrome OPS is great. First of all, you have a fair amount of horsepower in it to begin with, and typically those can last at least five to seven years. You have enough headroom and the horsepower to do that. Then you have a panel with it, and what's nice about the Chrome OPS is you could actually keep the panel for 10 years and you could update the compute very cheaply every five years. So you could get two computes out of one panel.
What you don't want, especially when you integrate the compute into the board itself, if the compute becomes old but the panel still works itself, you actually have to buy a new panel in addition to the compute. So your expenses go up dramatically. By having the Chrome OPS, which is kind of semi-integrated, you can continually upgrade that compute while maintaining an old panel. That's actually not a problem.
So it's a really nice way to manage your budget. You don't always have that expensive panel purchase. You can just buy a simple compute for it and the Chrome OPS is quite affordable.
Sometimes everyone gets a little obsessed with the front-end costs and it's really about thinking down the line and refreshing devices. And this sounds like it obviously helps that cycle along.
O'Malley: Yeah, you've got to look at the total cost of ownership for technology and upfront is one cost, you're correct. But if you look at the 10-year cost of a flat panel display, if the compute is integrated, it could be significantly more because maybe you need to upgrade in five years versus 10 years, and that's certainly going to jump the price significantly for a school district.
Even from a cost standpoint, you've got to look at if you do everything 100% cloud, what we start to realize is that your cloud connectivity costs can get pretty expensive pretty quickly. You start to get charged per student. And while early on it may not be huge, maybe you think it's good, but when you start to sum it up for a school district and you have 10,000 students, each of whom has a monthly cloud services charge or something, it very quickly becomes unaffordable.
It's a growing cost because the more you do with AI, the more you do with compute, if it's all in the cloud, your connectivity across is continually going up and up and up and up. And that's an area where after a short while, it becomes very practical to just build something on the Edge. So you can build enough capability on the Edge to manage it.
Then again, with a hybrid AI infrastructure, you can do 90% of your work on the Edge and then the 10% really valuable stuff you can offload to the cloud dramatically keeping your prices down. So that's something that school districts have the ability to do with edge as well.
How much of a challenge is it for a district to do that, in terms of resources and manpower?
O'Malley: I think building this hybrid AI infrastructure, there are some challenges there. Google's working on the same thing as well to build this for the educators who don't have to have the expertise to do it.
We'll make it such that every school district could change their requirements, but that's something we work with all of the educational content providers, the administrative software groups, all of these software vendors in this business. We will work with them to help with the structure of what goes to the cloud, what stays local, all of that.
Typical IT departments at most corporations can do it, but admittedly, sometimes educational IT departments aren't as big as corporate IT departments. So that's something I think the Googles of the world and some of the other software vendors, certainly Intel, we can step up to help build the functionality so it's configurable by the district in a manner that doesn't require ridiculous technology expertise to do. It's got to be user-friendly to do that.
How can teachers and students take advantage of Chromebox OPSs?
O'Malley: The key thing for a Chrome OPS, it's controlled by the teacher. So the teacher is going to be the prime user of it. They're going to be the one who figures how it works. They're going to be the one that drives the course content on it, and then they can allow the interactivity with the student. If it's a classroom where the students have Chromebooks, the teacher can set it up such that the Chromebooks and the Chrome OPS can constantly go back and forth. Students could share their screens to it, they could do quizlets, but it's very much a teacher-controlled device. It's not a student controlled device. It is a teacher controlled device.
The content that is on it is going to be teacher-controlled as well. That's actually an area where Google and educational content software vendors could work as well.
One of the challenges we've seen, especially with digital content, is maybe teachers spend too much time creating content for a class. What a teacher's good at is teaching. You don't want them to spend 30 hours creating content, then delivering it for 45, and then they have all the other homework afterwards. It makes it very unmanageable for a teacher.
Say you train the local Chrome OPS on the educational content, it can actually curate the content that you want to deliver. So maybe it shrinks the teacher's prep time from two hours to 20 minutes or 10 minutes or five minutes. And that's something that we've been working on. We're trying to figure out how to use AI or compute to help the teacher before class, during class, and after class. And the OPS can be an important role in all of that.
It allows the teacher to do what they're best at, which is teach. That's mostly what they're focused on. Mostly that's what they love to do, stand up in a classroom and actually teach the students. Let's take away all the administrative tasks and all the other things else and allow them to do what they want to do.
Anything else that maybe we haven't touched upon yet that you want to share?
O'Malley: I always like to clean up the idea that AI is not bad for students. It can be bad for students, but for the teacher, I think it's the greatest technology for education we've ever seen. If we can take advantage of that, it'll be phenomenal. And if we miss it, it's going to be bad.
There's been some data recently that students, particularly in developed nations, the education is declining somewhat, and I think AI use properly can change that. We have seen the greatest user of AI in teaching is actually China. If we can model a little bit of what they do, I think you're going to see the developed world, the education, start to rise again, which is certainly necessary in today's world.
There's no doubt students and teachers should be familiar with AI and how to use AI as a tool to help you get better, to do your job better,l to make you a better student, a better educator, all of that. I think it could be done that way, and that's what we need to search for in the industry to push for in the industry.
