WKAR is supporte by the MSU Research Foundation, bringing new innovations to th marketplace for global impact.

Learn more at MSU Foundation dot org.

Education is a very compli complex topic.

It's beyond just to teach our kids, it's more about training the next generation of our humankind.

Could I really use A.I.

systems to really enhance classrooms?

We are really trying to incorporate A.I.

in our classes, not only from the technical perspective, but also all the implications around them.

It could end up sort o magnifying societal disparities.

And that's what should be concerning all of us We have policies at the classroom level that say what you can use generative A.I.

for and what you can't use it for.

We need to leverage the power of A.I.

to make our classroom a better place for students to learn.

A.I.

is any technology that seeks to emulate human intelligence.

It's not one thing but it's kind of a suitcase word that describes a broad range of phenomena.

There are different types of AI.

On one hand we have predictive A.I.

that helps us.

For example, Forecast something.

If you're listening to songs on Spotify or you're using a social media platform, ordering a cab through Uber, all of these use predictive algorithms.

Then we have this new technology that has become very popular nowadays and that is generative A.I.

Systems like DALL-E and Midjourney create art and other outputs that you've seen in tools like ChatGPT So at the core of developing these A.I.

methods is obviously data.

It is the use of giant data sets and algorithms in order to produce a particular output.

An answer to a question.

I do think it's important to understand that this technology's not a search engine.

So it does not look through all the informatio and gather different websites.

It provides a probability that one word makes sense next to another word based on the inputs that you provided and the data that it has.

So the people who build the A.I.

play a very large role in determining what kinds of data the AI should pay attention to when it's trying to generate images, write poetry, You know, write poetry or do any of the other task that we might be interested in.

One growth area where A.I.

is being deployed is in educational systems.

The use of A.I.

in the field of education, is relatively new.

But there are really valuabl ways in which A.I.

can be used both to assist teachers and to assist children, to help them actually grow and develop their understanding deeper over time.

The goal of education is about mastery.

There are ways we can think about how we can leverage the power of A.I.

to build a more personalized learning, to really help students to master the skills that they need.

At a very high level, our A.I.

and we view it as a tutor for any student and a teaching assistant for any teacher.

The PH and PKA, that's wha we're going to be doing today, Typically, we have 25, 30, 35 students in a classroom.

Every teacher knows that thos students need different things.

But it's hard to do tha when you have 30 in a classroom.

I always give the exampl that if you were to rewind over 2000 years ago, Alexander the Great had Aristotle as his personal tutor.

And I would think that if young Alexander was having trouble with the concept, Aristotle would slow down a little bit.

If Alexander was unusually precocious, let's say military strategy, Aristotle would speed up a little bit.

Unfortunately, we don't have the resources to do that for every student.

A.I.

will be able to help teachers to do that, Because A.I.

is very good at using data t diagnose students' performance, to diagnose motivational status, and their engagement.

A.I.

is powerful enough to provide that diagnose on the fly.

In science education, we try to get students, try to write what's called a scientific explanation, which really includes making a claim between a variable and another variable.

What's the evidence?

And they have to support that claim.

And what's their reasoning?

This is a challenging enterprise to write one of these, and they're fairly long and in depth.

One of the things that A.I.

can do is support the teacher in giving kids feedback on their essay.

It's not giving them the right answer, it's giving them hints on how they can improve.

And they don't get it a week later, with A.I.

they can get it instantaneously and they can get good feedbac that could say something like, you know, I noticed yo mentioned one piece of evidence, but what other evidence might you have to really support this claim?

Are there other claims that also might be supporte by this data that you presented?

That's the kind of feedback could really improve their scientific reasoning and their understanding of what science is all about.

Researchers are looking into ways A.I.

tools could help personalize the learning experience.

Tailoring instruction for individual students.

Today, when we walk into a classroom, it's a very diverse group of students.

I have kids who might have come from a different country, who don't speak English necessarily as their first language.

Students can talk to the A.I.

They feel like they have an infinitely patient tutor always there to support them, a tuto that speaks multiple languages.

So even if the work is being done in English, they can get support in Brazilian Portuguese, or Spanish, or or some other language.

I have kids who come fro different socioeconomic status.

They have different world experiences.

True personalized learning would take all these factors in consideratio and then delivery instructions as differen from one student to the other.

We may bring some examples, stories that's relevant to them so they can better understand the content.

They can also relate to the content better to feel more motivated to learn in that content.

A.I.

can make suggestions to teachers about how they might be able to modify their lessons moving forward, reall to try to create that motivation about why science or other subject are really important, to learn by bringing up context that are more related to them Teachers are also looking to artificial intelligence to help manage large workloads.

The things that A.I.s can help us do although there's multiple areas, the augmentation of all of us in the jobs that we do and the tasks we perfor is the most exciting direction.

Teachers spend a lot of tim doing things like writing lesson plans, grading papers, writing progress reports, if the A.I.

can act as a teaching assistant and take some of those tasks, or at least shorten them and take them off of a teacher's plate, that's more time and energy for the teacher and more time and energy that the teacher has for the students.

A.I.

can handle a lot of the easy cases.

The teacher can handle the cases that are more challenging, particular, giving the teacher time to really work with students who might struggle.

The teacher can really go over and sit for a good, good 10 15 minutes to work with students to really try to help them push their understanding forward where the other students can be working in their group and maybe getting some feedback from the A.I., but they don't need th concentrated focus of a teacher, both in providing cognitive supports as well as motivational supports.

For the teachers t monitor what students have done.

We usually give teacher these dashboard looking things where they can look at the data and we still provide those.

But now they can talk to the AI and they can ask questions.

Hey, what have my students been up to?

Do you have recommendations for thing I might want to assign to them?

Can we use that informatio to inform what tomorrow's lesson plan is, which the A.I.

could also help the teacher draft.

Now, what's critical we're not removing the teacher.

The teacher is really critical in making decisions and setting the course for what's going on in the classroom.

More or less the A.I.

has become a partner to support the teacher.

But it's not replacing the teacher.

Many teachers and many schools, indeed are afraid that A.I.

in the future may replace their jobs, it's a very valid concern.

But the short answer is no, I don't believe A.I.

will replace teacher.

But teachers and schools do need to know.

We need to evolve with A.I.

If the schools and teachers are reluctant to change how they teach, to adapt, how they think learning would occur, they actually may be left behind.

From an educational standpoint we really as a society need to make sure that we don't create two different worlds, the haves and have nots.

We have to make sure that what's ever available all in one school system is also available in another school system.

Otherwise, we're going to continue to have this great divide it tends to go on by by the color of our skin, right.

So we have to make sure that within our society we have policies in place that make sure everybody has equal opportunity to use these technologies to promote their learning and that these technologies work in those various different kind of environments.

Otherwise we're just going to create different society, different classes of individuals and we have to do everything in our power to prevent that from happening Moving ahead has an educator, one thing that also concerns me is are we training students?

Do they have the algorithmic literacy skills to not only survive, but thrive in a world where everything is A.I.

generated?

What we're seeing is there's about 20% of students who, when they have access to, for lack of a better word, an AI tutor, they know what to do with it.

They ask the right questions, they engage.

And those students are off to the races.

it's not always the students who are always performing well to begin with.

Some of these students might have been struggling but they know how to use the A.I and they're really benefiting from it.

So one of the things we've learned is we need to figure out both in the product and maybe support and training for students and teachers so that more student can really start to understand how to use these tools.

There are growing fears that as A.I.

i incorporated into the classroom, students may become too reliant on artificial intelligence.

You know, we can look at the electronic calculator as an example.

We can misuse that tool and prevent kids at understanding how to add which is a real possibility.

On the other hand, when I'm doing work and I have to add u a whole series of large numbers, it's a tool that actually facilitates my work.

Like the calculator, A.I.

tools and other technolog tools can enhance what I can do.

So maybe we should better define the appropriate use of A.I.

in the school system as well.

Within universities, like here at Michigan State, we have policies at the classroom level that say what you can use generative A.I.

for and what you can't use it for.

I don't think it will make us dumber.

hopeful.

I'm a hopeful optimistic.

But here is the caveat to that.

I think it's based on how we use A.I.

So I think it will make us dumber if we simply use it to do everything for us.

To really use an A.I.

well, you have to be able to hang with the A.I.

You have to know what great looks like.

Someone who has a base of content knowledge, someone who's really willing to dig into their curiosity, A.I.

is going to be a superpower for them, while someone who's just trying to cut corners and and call it in, maybe it might help the kind of get through something, but broadly in life, they're going to be at a major disadvantage.

As the use of artificial intelligence in classrooms grows, so too does the concern around A.I.

's potential use for cheating.

As an educational researcher looking at cheating, I'm trying to find the fundamental cost, the root cost of cheating.

Is that a technology issue or is that an issu that's related to the student?

The reality is technology for all of human history, it's always amplified human intent.

A knife can be used to kill a knife can be used to prepare a fine food It can be used to build things.

Similarly, A.I.

will amplify human intent If someone has negative intent and there are people with negative intent, and I am worried about things like fraud and, deepfakes and authoritarian governments using AI for bad intent, it's going to be amplified.

If someone has lazy intent, they just want to not do the work, yes, there are ways to use A.I.

as there are ways to use other tools before it.

When you parcel out all the technology component, the fundamentals of cheating is student's motivation.

It's in the context.

It's in the situation.

It's in the environment that we create.

When we have performance oriented instruction, we focus on grades, we focus on comparing students with each other.

So when we talk about value in a performance driven type of curriculum the value of cheating is students needs to perform or appear to perform better, to make them appear more competent.

So if we really fundamentally wanted to address cheating behavior, we really need to think about curriculum reform instructional reform, and then to design better instructional activitie that are geared towards mastery.

As teachers, as educators, if we could create a desire for students to learn, they're not going to go to the A.I.

system and say, do this for me because they're goin to want to learn it themselves.

I have a few differen assignments that I have students do, and I think something that has really resonated with them in that they've been able to apply it to something that they're actually interested in.

So in my class they have to write a script for a new Netflix series.

They do actual script writin and they start to realize that if they just ChatGPT, give m a script about this broad topic, it comes up with very, very generic films.

So we have that conversation of how because A.I.

is trained on previous data it can help you brainstorm.

It can give you idea that a broad audience will like, but it won't give you an idea.

that is very unique.

So it really opens their eyes into, you know, I can use this technology, but I'm not going to be the best screenwriter to have it do all the work for me.

So instead of me telling them that, they realize it on their own, and that is, I think, something very, very good.

There's cases where we absolutely want to make sure that the student can write and do the research on their own, maybe supported a little bit, but even that should be transparent.

So what we've built into Khanmigo are a bunch of guard rails.

First of all, Khannmigo will not cheat for you.

It'll act as an ethical tutor.

It'll nudge you forward, it'll ask you Socratic questions.

Second, any interaction, especially if it's an under-18 student that they have with the A.I., it's transparent to the teacher If students try to do anything shady with the A.I., anything that sounds unproductive or unhealthy, parents, teachers, administrators get actively notified about that.

And then if we think about, say, use a case like writing, which I know a lot of folks are worried about in not just in K-12, but even in higher education, teachers are going to be able to construct assignments with the A.I., assign it through the A.I., and then the A.I.

will support the student, not do it for them.

Maybe.

Hey, this is the prompt.

What's your what's your opinion?

Let's come up with the thesis statement.

Then they'll do outlining together.

It's really like working on a document together.

It's not just a chat interface and it highlights parts of it and it nudges them forward, it gives them feedback on it.

But once again all the work is on the student.

It acts as an ethical coach and with the students ready, they can submit it to the teacher.

Now, what's new here is historically, when a student gave an essay to the teacher the teacher just gets the essay and they take it on a leap o faith that the student did it.

Although we know sometimes it was done with cheating.

Well, before A.I.

Now our A.I.

can make transparent the entire process to the teacher.

Sal and I spent fou and a half hours on this essay.

He had a little bit of trouble coming up with a thesis statement, but with a few questions, and I was able to nudge him forward.

Here's our whole transcript of this conversation as we worked on the paper.

By the way, this essay is consistent with Sal's other work, so I feel confident that it's from him.

If I went to ChatGPT or I went to one of these services on the Internet, thatll write my essay for me Khanmigo is going to tell the teacher, I don't know where this essay came from.

We didn't work on this together.

And it's not consistent with Sal's other writing.

So not only will it undermine I think, all forms of cheating, but it will also support students and teachers a lot better, which I think is the most important part.

The data that goes into the A.I.

models is critically important to avoid complicated issues like bias being incorporated into the systems.

You have to remember A.I.

is literally a dumb piece of technology.

We put all the intelligence into the system.

A.I.

is based on data sets.

Those data sets are extracted from our world.

We live in and have historically lived in a pretty unequal and unjust world.

And therefore the data that A.I.

is based on is unequal and unjust.

The issue of bias in A.I.

algorithms is something that the research community is aware of and I think a lot of peopl are taking this very seriously.

We want to solve it.

People don't want to build systems that are discriminatory or biased or hurt people.

From a technical standpoint, how do we build an A.I., especially for kids and schools that is not biased that is scientifically driven, that is also, educational oriented.

There are two ends that we can gear towards.

One is to produce a more sound A.I.

algorithm to reduce bias from a technology standpoint, but at the same time, from a consumer standpoint and maybe we want to be mindful about what A.I.

outputs could be bias in the output of A.I.

People have talked about making sur that people who are doing that procurement are trained i asking the critical questions.

What is the data that this is based on?

What did the vendor do to make sure that any biases in those data sets were corrected?

In areas where there are, you know, sort of long standing questions about inequity and injustice, you might ask, to what extent are the users of these algorithms trained in implicit biases and other kinds of, structural inequality so that they know about both the potential problems of the algorithm, but also the potential problems with their own interpretations of those algorithms.

We do need to trai our teachers to also understand, A.I.

and understand the bias involved in A.I.

so they can better prepar our students to deal with A.I.

At the same time we can directly train our students about those topics as well as our parents.

So it takes a village really to achiev and then to help our students.

What can w do to better prepare ourselves and our students for the transformation that artificial intelligence is bringing to education?

I tell everyone, take a breath.

I tell that to myself, meditate.

You know, everyone is feeling this.

It's not like someon has some massive advantage here.

The fact if you're watching this documentary, if you're thinking about these things, you're already on the leading edge.

But I think it is important for folks to also not be complacent about it.

Whatever is coming out, try i out, form an opinion about it.

Think about how to use these in your own life, how you can use it to suppor others in education or whatever, whatever domain you're in.

We all have access to these systems.

If they're going to have access to this, I want my students to be productive consumers of information and critical consumers of information.

So if instead of “don't use it because you might plagiarize,” use it in a smart way, and be able to use it in a way that it actually enhances their learning.

What the existence of these tools has brought forward, I think, for us is that we actually have to think through some of these challenging ethical questions, and it's going to take a lot of partnership between, people who build the core technologies, folks like myself in computer sciences, but also people who think deeply about...about ethics.

And also for the broader society, because these are questions that need to be answered not just within, the hallowed halls of a university or within Congress, but they need to be discussed with the public.

The public has to weigh in on these things as well and figure out what are the right ways to behave.

A few years ago nobody was thinking about tech and artificial intelligence.

But now I'm having conversations about it all the time.

It's on the front pages of the newspapers.

We have students from computer science and engineering, as well as public policy, saying that they want to help think about how we can engage in more responsible development of technology, how we can think about different kinds of policy levers to make sure that A.I.

actually serves public good.

So what does the future hold for artificial intelligence in education?

Its probably going to shock us what A.I.

will be capable of in five years, probably even in a year or two.

Youre going to have the intersection of A.I.

and other technologies, things like augmented reality and virtual reality, and my best guess is those will start getting mainstream in about five years.

The devices will become low cost and light enough to use them.

And then you can imagine you put that on and you have a tutor that appears before you and they can take you to ancient Rome and be transported into the circulatory system and that sounds like science fiction.

I used to think that the holodeck on Star Trek was the least realistic part of it, but we will probably have that in the next five years.

Right now, when we interact with an A.I.

system, we're interacting with text or just images, but the way that we experience reality is multimodal.

We hear, we feel, we smell, we touch and the A.I.

systems of the future, I think will also be able to interact across all of these modalitie at once.

That's going to unlock some really exciting opportunities, things I'm personally really excited about for the future.

We need to leverage the power of A.I.

and we need to build into our school system to better train our students, also provide learning opportunity for teachers to learn about A.I., to gain that A.I.

literacy to make teachers a better teacher, to make our classroom a better place for students to learn.

We do need to be more purposeful in terms of planning A.I.

and technology integratio and then provide resources, time and training to the teachers so that they can do this.

This technology is going to keep growing.

So we don't want to just teach our student how the technology is right now But as the technology evolves, how can they be critical thinkers of that technology and innovate the ways that they will be doing?

Because at the end of the day, we do want them to graduate and be the leaders.

I think it is possible that we can incorporate it into our lives in ways that we've incorporated previous labor saving technologies.

Its not likely to take over the world and, you know, start to make decisions on our behalf.

Certainly not in five years, and I am very skeptical.

My hope, is that w can find a way to live with A.I.

that minimizes the harms but maximizes the benefits and takes its potential to make our lives easier seriously.

The explosion of A.I.

and generative A.I.

has really cranked up the excitement that exists in tryin to learn about how these tools can actually be used productively to make our life better.

How do we get everybody in the world to have sustainable food?

How can we ensure everybody has clean water?

It's not just climate change.

There are other topics within our world, within our society that learners are very interested in, and that's wha we have to drive our learning.

It's learning about important thing that our kids are interested in.

To me, a good measure, if we're successful, is have we actually create the joy of learning once again?

WKAR is supporte by the MSU Research Foundation, bringing new innovations to th marketplace for global impact.

Learn more at MSU Foundation dot org.