What AI Frees Us Humans to Do

Photo by Eric Krull on Unsplash

I spent last weekend in New York at the 4th Neumann Series Conference hosted at Columbia University. This year’s theme was “The New Frontier: Human Ingenuity in an AI-Driven World” (here’s a link for anyone interested: link). It’s a wonderful series, is free and open to the public and privately funded by people who care about science and free discussion. I’m very proud to be involved.

The series brings together top scientists from around the world to discuss topics broadly under the theme of things inspired by John von Neumann (1903-1957) who was a true genius and polymath. It’s hard to explain his contributions to science and the modern world because they are so great and so fundamental. Saying “he founded” or “contributed to” a field doesn’t quite capture it. Google him on your own. Suffice it here to say that he was fundamental to the fields of mathematics, physics, game theory, computer science, and AI.

Late in his life he penned an essay entitled “Can We Survive Technology?” As the mathematician behind Los Alamos and the nuclear bomb, one can understand his late-in-life question. The essay, however, is more about technologies - many of which he created - evolving faster than humans and human institutions and the challenges thereof. It was the perfect launching point for discussions about AI and its role in sciences, research, education, and our lives overall.

For those who read my earlier column criticizing “AI Insanity” - which is what I call the hyperventilation around the idea that AI will replace all humans in everything - you won’t be surprised that I’m a bit of a realistic-optimist. I understand and expect serious challenges to the current world order, our jobs, and my job in particular as a researcher and teacher, but I see the net gains as tremendous and the most dire, dystopian predictions as unrealistic.

Much of the concern around AI is actually around AI replacing humans and thus around the implied devaluation of humans. What I didn’t expect from the conference was that I would become deeply convinced of how much AI is moving us in the opposite direction, further raising the value of humans and human interaction. That’s the topic of today’s column.

What follows are some of the relevant comments from the conference and how I pieced together my conclusion.

Robots Lifting Weights in Physics Class

One of our hosts was Columbia Physics Professor Szabolcs Marka. He began in the morning by explaining that he’s optimistic about using AI in class and that he encourages his students to use it. But, he explained, he tells them to use it to help them learn instead of do work for them: “I tell them that it’s like having a work-out robot to help you get in shape but instead you have it go to the gym and lift weights for you. How you use it is totally up to you, but one way benefits you and the other way does not.”

The point then is that all of us should use AI to do the lower-level tasks we understand and essentially delegate to AI but not for new tasks we don’t understand yet, unless, of course, you are asking it to teach you about that new task so you will understand it. In my principles of macroeconomics course, I tell my students what prompts I used in ChatGPT and what notes I wrote versus what I had Chat write for me.

When asked how AI affects him and his grad students or fellow research partners and what tasks they assign, Prof. Marka commented that AI has been great. “We can spend more time now walking, talking and thinking. Then when we have an idea we think is worthwhile, we can have AI quickly start on some initial tasks and calculations. In the past that might have taken us 6 months to work through. Today we do it 1-2 weeks. It means we can make progress on problems much more quickly and thus spend more time together discussing exciting ideas and new information!”

More One-On-One Time to Play

This idea that AI sort of “flips” the classroom so that time with the professor is spent in discussion rather than lectures where information only flows one way (from professor to student) is not new. We saw this with the advent of online courses, books, testing and teaching materials and so on about 15 years ago. The thinking at the time was that students could learn material online and then come to class ready to discuss. The teacher would be more like a discussion leader, engaging eager minds in an open forum reminiscent of ancient Greece or something.

The panel discussion on Learning Reimagined addressed just this. Imre Kökényesi is an inventor-entrepreneur, mathematician and physicist. He explained that he focuses developing games that teach fundamental mathematics. He believes that mathematical reasoning is based on only a few fundamentals, can be taught in early childhood via games and, once learned, mathematical reasoning will open young minds to unlimited possibilities. So, he has developed games and pedagogical material for early childhood learning that will be tested in a few pilot schools this year already.

He explained that this returns us to learning more one-on-one like we learned from our parents and the way others learned earlier on in history, in his opinion. We learn better when we are engaged with a human. It’s not possible to teach well, in his view, in a class of 30 students, it’s much better to have one teacher and a few students. The teacher-student the time should be spent “playing” and learning rather than lecturing. He discussed how AI is helping this come about.

Myriam Da Silva runs a company providing AI-based educational platforms that allow just that. Again, she discussed how this new technology frees the teacher to spend less time “lecturing” and more time engaging with students. This will, or should, in her opinion, also mean smaller classrooms so one teacher can more meaningfully engage with those students.

AI here is being employed to increase human-to-human time. Notice, the classroom got smaller. That’s more teachers per student, not fewer. That’s the opposite of what we have normally been worrying about: AI replacing teachers.

But, I think we can all see from this discussion that, if we indeed move in that direction - and clearly its the one we all want to move in, whether it’s feasible or not is another question - then we’ll likely also reshape how education is structured. Maybe school only needs to be half as long since learning from AI can take place anywhere. Then, the time we do spend in school should be more intense and impactful. Maybe, as a professor, I only have one meeting a week with my class of students and otherwise meet them one-on-one. I don’t know. But it’s clear that the current structures and institutions were all built for a different kind of learning. AI is freeing us to spend more time together, not less. And because that “human time” is more valuable than “AI time”, we want it to be quality time.

Doctors, Patients and AI

Two sessions included panelists that touched on aspects of AI involving medical or psychological practice. Nóra Árvai, an applied health psychologist at the University of Debrecen, has launched a consulting business that helps doctors work with patients that are coming to them with more developed opinions and questions than in the past due to greater information availability and, today, AI.

Dr. János Réthelyi, a psychiatrist and researcher at the Semmelweis University, discussed how patients are turning first to AI like ChatGPT for basic therapy questions. Whether this is good or bad is an open question, but, regardless, psychiatrists are learning to deal with it.

Two other speakers hitting on the same issues were Lili Gerendás, a PhD student at Cornell Medical School, and integration engineer Rebeka Bojboi Van Batenburg works at Zocdoc where she works on integrations that connect healthcare providers, scheduling systems, and patient access platforms, contributing to more efficient digital healthcare infrastructure.

In all these cases, the panelists explained that AI gives good advice and bad. People need to be careful. And they explained that AI is rapidly changing their jobs and, specifically, interactions with patients. AI is helping patients with basic questions and helping doctors with routing tasks of notetaking, filling out endless forms, conducting basic analysis, and even reviewing imaging data which it now does in some areas far better than humans can. While there are challenges, to be sure, all of them concluded that the direction in which they are moving is toward AI helping with basics and freeing doctors to spend more quality time with patients.

The more things change, the more they are the same.

All this modern AI talk has only allowed us, once again, to ask the fundamental questions thinking people have been asking since the dawn of time. Certainly as least since the days of Socrates and Plato… why are we here, what’s our role in society, what makes society “good” and how do we define that? And so on.

Return to Prof. Marka’s comments. AI can teach his students mathematical techniques to solve high-level physics problems. AI, however, can solve the problems itself - hence his warning about using the robot to actually lift the weights - or teach the material - actually doing his job or, to continue the analogy, lift weights for him. We know this. It raises a very different question: Why do the students need him?

What is his role as “physics professor”? What is he teaching? Techniques? Topics? Answers?

I posed the same question to myself about 15 years ago when textbook companies launched online platforms with their books. Suddenly the textbook, which was written by a famous economist, included homework and exam materials and a platform where students could do the problems, get them graded and get grades. The platform provided a tutor so that students could get help in the form of links to the location in the chapter where material is covered, additional questions to help them learn, videos of the author explaining the topic again and, eventually, a collection of other professors teaching the material in videos, digital platforms and so on.

I was asked to develop an online statistics course at the time, which I did. I started by asking myself: “If my students can get all of that online, what value do I provide?” I can edit the material covered to ensure it fits in a semester and then just sit back and let it run. But why? Why put my name on the course at all? What’s the point?

The answer I came up with is that my value is in helping explain the harder parts, the stuff that is hard to learn from a book. We’ve all had the experience trying to learn some topic you can read about until you are blue in the face but you still don’t fully “get” in a deep sense. Then you see some person explain or demonstrate it and voilà suddenly it all makes sense. My job, I decided at the time, was to know those pain points and help them through it.

I ask the same thing today in my class. What is my role? Why are 30 students physically sitting in seats? And why did they choose to be here? They could easily have taken this same intro economics class online. It would even be an online version of the class developed by our faculty for our students. Nevertheless, these students signed up for my class and physically came in to sit and learn. Why?

Well, my students tell me this: Today, I take online versions of the classes I don’t care about. I know if it’s all online that I don’t have to learn the material and I can get an A+ in the class. So, I focus on taking classes in person for stuff I care about or for material I really need to learn.

That, in my opinion, is the heart of the matter. And it’s true of AI now in exactly the same way it’s true of all the online materials students have available. And, my prediction is that it will be exactly the same. For things I don’t care as much about, I’ll have AI do it. That will then free me up to do things I do care about.

And that’s why it always brings us back to the basics, forcing us to ask: What should I care about? Why am I here… to spend time on unvaluable things or on valuable things?

All the conversations at the Neumann conference drove this point home to me. None of the speakers spoke about it directly. They focused on the role of AI, how it’s changing their research, their field and so on. And, just so my column here is not putting words in their mouths, to be clear, many of them, about a third of the participants, based on a show-of-hands poll, felt AI will be a net negative. But that is not what I heard and distilled from the discussions. I drew the opposite insight, even from the skeptics.

Every single example from the physics professors to the medical innovator to the psychologist to the artist: AI was being used to (a) serve a human and a human end (see my older column) and (b) to free them to focus more on what really mattered and that generally was engaging with other humans.

By delegating routine tasks to AI, including basic analytical tasks that need to be performed, we are all left asking what is the role of the human, what is our role as doctor, psychiatrist, and teacher. These are not new questions.

Certainly we must still know our subject matter, but perhaps we will no longer be expected to be walking encyclopedias with perfect recall. In my classes today I have my students look up, say, the current inflation rate or unemployment rate and we use that in our example on the board. Sometimes I ask them to ask ChatGPT to explain a concept and then we discuss it. It doesn’t diminish my role as their professor. It enhances it. It hones in on where I contribute to their learning.

And… the interesting takeaway that I want to share with you all is that, again, in all these cases, yes, AI is changing things and eliminating certain jobs, but the clear direction is that AI is freeing people up to spend more time together, not less. It’s showing that we might be moving to a world with more teachers per student and more meaningful one-on-one with doctors, not less.

John von Neumann and the Computer

The conference ended with a short documentary on the life of John von Neumann. If can get online access, I’ll share it. It’s very worth watching.

But, for our purposes, I will mention that the documentary drove home one final point for me. John von Neumann was the mathematician behind the atom bomb’s development in WWII. He invented the modern stored-memory computer in part so that he could do more calculations and do them more rapidly given the urgency of the project. His wife became the first programmer for this sort of thing and wrote many of the first modern computer codes.

We need to remember that this was a time before computers, before calculators or even electric typewriters. Businesses and government had “armies” of typists in rooms working on documents and “armies” of people working on basic calculations for all sorts of things.

Von Neumann explained that the computer he invented would, in his words, “easily replace ten thousand people who currently make these calculations”. And he was right. The computer eventually eliminated all those jobs and all the jobs of those typist jobs too. AI is doing the same today.

Today it’s not surprising to hear that someone is studying “computer science” but before the invention of the computer it would have sounded more like science fiction. Could you even explain in an understandable way what facetiming, direct messaging, engaging in social media and all the normalities of modern life are like to someone from von Neumann’s time?

Will we one day have “AI science” that is doing things we can’t even conceive of? The answer is “yes”, whether it’s called AI Science or not. Will there be jobs we can’t even imagine today? Yes.

Conclusion

My students say it all. I take the classes I care about in person and take online classes that I don’t care about (so I can get an A without learning anything). AI is no different.

The valuable piece is the human piece. It’s the class you take in person. The automated piece gets devalued, not the other way around.

Will this change things? Yes. And that is both exciting and scary, I readily admit.

It also puts a higher burden on the in-person, on the human piece. If my students take my class in person, it better be good. I better engage them and teach them. I better know my stuff and know how to help them.

AI is taking over the mundane. That will free us up to do other things. And we repeatedly reveal that we value the personal, the interpersonal, the experience. We all need to think what it really is we do. How do and how should we spend our time on this earth?

I’ll end with wise words from my mother, Sharon Ball, who founded a not-for-profit at-home hospice in the early 1990s. She often reminds me that in all her years tending to people in their final moments, not a single one of them on their death bed wished they spent more time in the office. They always wished they’d spent more time with family, friends, and the people they love.

AI is freeing up our time. It’s up to us what we do with that.

May we take my mother’s words to heart and may AI help free us up to spend our time more valuably.

Thank you for reading.