- Writing and creative expression are frequent topics on this show.

But today's guest challenges all of us to consider the joy, the practice, and even the challenge of reading, especially in a world dominated by artificial intelligence.

She's Naomi S. Baron this week on Story in the Public Square.

(uplifting music) Hello, and welcome to Story in the Public Square, where storytelling meets public affairs.

I'm Jim Ludes from The Pell Center at Salve Regina University.

And our guest this week is Dr.

Naomi S. Baron, Professor Emerita of Linguistics at American University.

She's also the author of an important new book, "Reader Bot: What Happens When AI Reads and Why It Matters."

She joins us today from suburban Washington DC.

Naomi, it's lovely to be with you again.

- It's a pleasure to be in the ocean state.

- Well, you know, we had a great conversation with you back in 2023 on your last book, "Who Wrote This" was the title of the book.

And you were looking there about whether, what happens when AI writes.

This new book is looking at what happens when AI does the reading for us, or what's the appropriate use of AI reading for us.

But what draws you, as a linguistics professor, to these big, broad questions about AI in our writing and in our reading?

- What draws me is I care about human language and our not losing our skills for the reading and writing thereof.

And as a person who's been a university professor for a million years and worried about whether students believe it's worthwhile doing your own writing, not buying a paper from somebody else.

And whether it's worthwhile these days to do your own reading when there's a tool available, you know, on many different kinds of large language model sites that can quote "read", and we can talk about what that means in a minute, for you.

And you don't need to do the reading in order to produce the writing that you then submit as reviews of literature or analysis of Herman Melville's work or reviews of scientific articles or whatever your field happens to be.

What happens when we humans no longer feel an imperative to be able to do that reading and processing and analyzing and developing a certain point of view of our own on our own?

- So let's get right in, then, to Reader Bot, because there are lots of different threads that I wanna pick up, but I think it might start with, if we're talking about reading, this sounds like a ridiculous question, but what is reading?

- For human beings?

- Yes, for human beings.

And then we'll talk about what it means for the machines.

- Okay.

When children, or sometimes adults, learn to read for the first time, we talk about learning to read, meaning learning how to decipher words on a page, and that itself can be a fun process, a tortuous process, but you're basically trying to make correspondences between what the symbols represent.

It could be an alphabet, it could be a syllabic system it could be a character system, but making a correspondence between what that written symbol is and what word it's representing.

And then if you're lucky, you progress to a stage of reading to learn, where the goal now that you've processed how to do decipherment, where the goal now is to get new kinds of ideas and thoughts and emotions and experiences that other people have had and written about into your own head.

There's all kinds of literature on the psychology of reading and how the eyes move forward in forward circades and backwards circades, and then going to the next line and so forth.

But I think what's most important to know when we talk about what it means to read is taking, as a goal, an experience that you make your own.

And that experience could be for learning something new.

It could be for sheer pleasure.

It could be to de-stress.

It could be for 1,000 different reasons.

What worries me, to jump ahead to Reader Bot, what worries me isn't even so much what happens when the analytical component of reading is outsourced to AI, what in some sense matters to me even more is the reasons that we read for ourselves, whether it's for pleasure, whether it's for learning from the experience of a character in a novel, this is what the so-called "Bildungsroman" or coming-to-age novel, coming-of-age novel was all about, of seeing how people in various life circumstances have handled those circumstances, how they've transformed themselves as people.

And you ask yourself then, if you're a mindful reader, "What does this have to do with me?

What experiences have I had that are similar?

How can I live a different life as a result of having done this reading?"

AI is not poised to do that kind of reading for us.

My concern is that as we use AI less for the business of life purposes, that we will no longer care to spend our time, and effort sometimes, reading for becoming different people.

- I worry, as an educator, I worry that my students don't appreciate the value of a book.

Just over the years, if I don't assign, they have to have their papers have to have actual bound books in their bibliography, all I get are, you know, Google sources and popular accounts.

But even beyond that, I think about my own experience with books, which have been transformational.

I'm 13 years old, I have the chicken pox, my dad brings me a copy of "The Killer Angels" and it opens up a world of study and curiosity and ultimately careers because of a book.

AI's not gonna do that for us if it gives us a summary of three days of the Battle of Gettysburg.

- All right, but we have to be honest with ourselves.

The decline in reading done on a voluntary basis has been going on for quite some time.

In different countries, the amount of decline is not always the same, but it's documented now, for sure, in most English-speaking countries at least in Western Europe and in the United States.

And that happened long before what I like to call modern AI, which basically was the development of the transformer model, which became the basis for OpenAI, creating what we now think of as large language models.

They existed before OpenAI, but, actually about the same time.

But what happened was... We came to have a new notion of how we were going to use artificial intelligence.

And first it was for writing with the coming of ChatGPT, and then we began to recognize that a-ha, in order for AI to write, it has to do some version of reading.

Oh, you know what?

I could set AI to do more reading for me.

And it's only within the last two years, basically, that you started to see students recognizing, students in particular, but also researchers, that, "You know what, I don't have to read all those scientific articles."

I'll focus on the researchers for a moment.

I could get them summarized, not just with the abstract that often comes with these articles, but with a larger, broader summary, and I could get, I could get the articles selected for me from this huge database, then I could get them summarized, I could get them compared, and that would hasten my work, that would save me time, and it sounds like a productive thing to do, but it's only really within the last couple of years, first within the research community, I think that's fair to say, but then more recently among students that there was the recognition, "Guess what?

I don't have to do all this work."

It's not just the writing of, you know, create an essay for me, but really read the materials first.

And then create a much more interesting essay.

- You know, but at the core of that is, it seems to me, a refusal to linger in the work and to do the labor of writing.

There's a reason that scholars are expected to write literature reviews, and that there's a reason that they're expected to have mastered that existing literature, not just sort of, this is some, a box I have to check, and so if I've got a summary, that's sufficient.

And I don't know that I've got a question.

It just maybe might be my own pet peeve.

But the question I wanted to ask though is, does AI actually read?

What's the machine doing when it's so, you know, supposedly reading?

- Predicting.

So predicting what?

There is a very large data set, some of it legally acquired, a lot of it not, that is used by the big tech AI companies to take basically now billions of words and break them into what are called tokens, which are similar to words, but not exactly the same.

The technology doesn't particularly matter for our conversation.

So we'll just call them words for simplicity.

And then you have algorithms that are part of that large language model that make a prediction as to what the next most likely word would be in a sentence.

And that is the way in which the writing that we see generated by AI comes out, by predicting word by word as to what it is you might want to say if you're asked to write an essay on Macbeth.

Okay, now let's say you are asked to, for your school assignment, to compare Macbeth with, "Oh, what would you like?

Would you like Lear?

Would you, you know, pick a favorite."

- Sure.

- Or do you like Sophocles?

You know, we could go to he Greek tragedies, as well.

And you could take Oedipus.

So then what happens is both of those books or works are, quotes, "read" and then the prediction is how a reading of the two of them, in light of the prompt that you gave, what predictably would be the next most likely word for writing the analysis that you, let's say the student, have been asked to write.

The really creepy thing is these tools have gotten incredibly good.

There used to be a lot of talk and turn it in used to be a monitor tool, and then there are various other kinds of tools, you know, like ChatGPT like GPTZero trying to say, "Can I prove that you have used this tool to do the generation of the work that you, student, were supposed to do yourself?"

It's pretty clear that it's become almost impossible to tell the difference.

There are enough studies that have been done where people are darn sure it was AI that did the writing, but it was a human and vice versa.

And as the tools get more and more and more powerful, the work that is generated is many times far better than what even a graduate student could produce.

So where do we go from here?

It's a values question.

The same as it's a values question on whether you read literature yourself for self-betterment or for enjoyment.

It's a values question on whether we believe that exercising our brains, let's say, for so-called critical thinking, a term I hate, but it's used so widely, people say, "Okay, I know what you're talking about."

You know, it's reading something and stopping and pausing and asking, "Well, what does that mean?

Do I agree with it?

Do I disagree?

How does this relate to personal experiences I've had and therefore going forward as I read, how much is my interpretation of a character clouded by someone I know, as opposed to what the author had in mind for the character?"

And you can play all kinds of combinations here, but the question is whether, to go back to education, whether we as educators are sufficiently convinced ourselves and therefore can convince our students that these kinds of mental activities are worth it.

One question to ask, and we'll just ask it hypothetically, is how many faculty members regularly read of their own volition?

Not because their job told them they have to.

And I think we may be sadly surprised to find the answer is not as large as one would like to believe.

So if they don't have the value themselves, don't watch YouTube, don't watch Netflix, don't just flip through whatever on your phone, but sit and read a book just because you can, and it's a wonderful alternative to some of the things that most of us are spending too much of our time doing.

If they're not feeling it's valuable to do it, can they, with a straight face, tell their students it's valuable?

- It's a great point.

I had an experience this year in my own classroom where a student, I let students have open notes on quizzes, and a student had forgotten to print out her notes.

And so when she brought them in, I printed them for her and I, it was 12 pages of typed notes for 24 pages of reading.

And I thought, "That seems like a lot, but okay."

And what I realized as she sat there struggling to answer five short questions was that she had taken the reading, run it through an AI and generated notes, but she had still not yet grappled with the content.

It was still, it was just another 12 pages that she had not really been exposed to yet.

And so I guess what I wanna ask is, if we rely on the machines, on AI, to do the reading for us, what do we lose?

- To answer that question, I will go back to the origins of literacy and ask, why did humans in various parts of the world, probably the earliest having been around 5500 BC in Mesopotamia sorry, 3500 BC.

I was adding the years for the current centuries, the current millennia.

Why 5500 years ago, did people feel having a system that gave a durable representation of things in your head, why did they feel it was worth it?

And then you go through the ages and you look at the use of writing for law.

You look at the use of writing for religion.

You look at the use of writing, later, for literature.

Then you look at literacy movements.

Again, pick your different parts of the world.

Why has it been deemed so important for people to become literate?

What is it that we thought was happening by gaining literacy?

How did it mold us as individuals?

How did it mold us as members of social groups?

If you have a religious text that brings together the members of that, of the believers into a community in a way that wouldn't happen if you didn't have written texts.

We know, to jump ahead a little bit, from studies that have been done, that our brains really do change neurologically when we learn to read.

What does this mean?

Well, you know, interpreting that a leap to the neuro psychologists, but we know that connections get built up of a sort that don't exist when you're not literate.

And then the question becomes, what are those connections, perhaps, potentially good for besides reading?

Are they good for thinking?

Are they good for exercising your brain in ways that you wouldn't have otherwise?

Now, it's very important to bring in that you can be in a non-literate society, and there have been many of them over the millennia, where there's some really smart people.

We don't know too much about what they have thought because it hasn't been recorded, but anthropologists will be quick to point out among non-literate populations of which some societies, of which some still exist today, there's a lot of wisdom.

So I don't want in any way to undercut intelligence issues by whether or not one is literate, but we know additional things happen as a result of being literate.

So what happens if people today are saying, "Ah, why am I spending this time?

I'll put things into an AI and then to amuse myself, I'll go watch a Netflix movie."

What have we lost?

Well, we've lost asking ourselves, what are all the things literacy has done over the millennia and are some of those things we will miss, we will miss emotionally, we will miss cognitively?

- Well, I would imagine also consequences for broader society, right?

If we think about a piece of literature challenging us to be empathetic or making us consider the perspective of the villain per se, those are things that we lose, but I wonder if we come back to higher ed for a second, this seems to be an existential moment for higher education as we've known it.

You spent a lot of years in the classroom, you've been a university administrator.

What do you think AI portends for the future of the academy?

- The question is complex because at the same time that what I'm calling modern AI has hit the big screen meaning people in general around the world are getting access to it, other things have been happening in higher education.

So the kinds of challenges that higher education has are not limited to what challenges AI is introducing.

So we've had, particularly in the United States, the cost factor, we've had the political factor with political parties to remain nameless statistically saying, particularly liberal arts education is not worth it.

Interestingly, there were studies done a couple of years ago where many people from party X believed that liberal arts education meant you had to be politically liberal to want it.

Oh, no, that was not what liberal arts education was about.

Liberal arts education was started with the trivium and the quadrivium in the Middle Ages long before there were American political parties, much less America in the way that we had westernized it to be.

But I think we have to ask ourselves, what is it-- What is as educational institutions we have as our goals?

So many institutions, including mine, are putting increasing emphasis on how you can get a job from your liberal arts education.

And I understand the reasons for doing that, but I reflect back on my own undergraduate experience and my time when I was teaching at Brown for many years.

And the saying at Brown was, you don't go to college to make a living or to learn to make a living, you go to college to learn to live.

And that notion of going to college to learn to live is being sidelined long before modern AI came on the scene because there were financial challenges.

There were demographic challenges.

There aren't enough would-be 18 year olds who want to go to college these days.

How are we gonna get them to want to come?

How are we gonna say, you know, that there's a return on investment?

So all of that came into play even without considering AI.

So now let's come to what AI is doing to higher education.

Indeed, it's making us rethink, particularly 2023, 2024, there was a huge amount of discussion, of chatter, if you will, particularly among teachers of writing or anyone teaching a course that involved composition and essays and so forth.

There was a lot of discussion of the need to rethink what kind of writing assignments are meaningful.

And the usual answer was craft those assignments in such a way that they're sufficiently personalized to an individual so that AI could not produce a meaningful essay for the assignment that you, the professor, have given.

- Well, guess what?

- That was my standard too.

- But you can standard, you can personalize your chatbot.

- Yeah.

- These days, I mean, people in the business world are doing this.

My son who's an economist knows how to tailor his use of a particular LLM so that it can draft emails for him in his writing style.

And he could write, you know, he could say where he's from and how, you know, whatever he wants to add in to personalize.

So that's probably not gonna work.

The reason that so many, comparatively speaking, from zero now to some number of faculty members are saying, "You're doing the writing here in class."

The reason they're doing that, and no devices are allowed in the classroom.

You need a pen, thank you, and some paper.

And the reason they're doing that is in part to prevent cheating, but sometimes, and this is the better side of it, to say sitting and writing by hand is a good experience.

I've done research this was for my previous book on "Who Wrote This?"

asking people what they like, what they don't like about writing by hand versus writing on a computer.

And there were so many people who said, "Yes, writing a computer is faster, and it cleans up my spelling and grammar when I have the right tools on.

But writing by hand is me, it's personal.

It's an expression of my thoughts in a way that when I'm writing on a computer, it's not me."

Okay, so students recognize this difference.

We have to teach them how to write and write cursive because it's faster, right?

- Which they're no longer teaching in most schools in most parts of the United States.

But students get it, and they're not all looking to cheat.

They're looking to ask, "What is it you're asking of me?

And if you're asking me to perform a particular way that's not meaningful as an exercise to my brain, who cares what I do?

You don't care.

You're probably grading my essay with AI anyway, so why don't I just finish the assignment, we'll both be happy, and we'll move on."

So one has to find motivation.

Similarly, what a number of people are doing, university faculty are doing now that they recognize students are not doing their own reading, is asking students to bring their books, print preferably, and to read together in class, to read aloud in class.

Now, that may sound ridiculous in a graduate seminar, but maybe not, because there's a level of engagement.

And from research, not just I, but so many of my colleagues have done of reading and print versus reading digitally, we know you read more mindfully, you remember more, you tend to pause more and think about what you wrote, think about what you're reading rather.

So there are ways to valorize both reading and writing, and I think that's the major transformation, not figuring out how do I make an assignment where AI can't do it for you.

But rather, how do I make reading and writing be activities that are sufficiently meaningful to you that you won't particularly want to use AI tools to do the work for you?

- Naomi Baron, this is a hugely important set of issues.

Your book is "Reader Bot", it's a must-read.

Thank you so much for being with us this week.

That is all the time we have, but if you wanna know more about Story in the Public Square, you can find us on social media or visit salve.edu/pellcenter where you can always catch up on previous episodes.

Thank you for spending some time with us this week.

I'm Jim Ludes asking you to join me again next time for more Story in the Public Square.

(uplifting music) (uplifting music continues) (uplifting music continues) (uplifting music continues) (upbeat music) (no audio)