🎹 Music for this post: https://www.youtube.com/watch?v=tPjoWN0SCb0.
Is a written piece inherently valuable?
Does the world need more writing?
Does it need more writers?
Or would it benefit from more original thought?
While I am not exactly mesmerized by ChatGPT, I do enjoy it as much as any new toy I’ve had in my hands throughout my life. There is no doubt that it can — and, likely, will — have a significant and positive role in the development of our civilization. I am aware that this is at odds with much of what is being written of late, so if you choose to proceed reading, I appreciate your willingness.
I am thankful for the public discourse that all manner of generative AI has spurred in the last five months, but as with all major shifts, it is amusing to watch people struggling to keep things in historic perspective. As with Clever Hans or many other magic tricks, it’s wise for onlookers to get a grip on the reality behind the illusion. Generative AI is merely the world’s most advanced parrot, underpinned by an ingenious application of statistics. If you haven’t read that last link (courtesy of Stephen Wolfram), you owe it to yourself, because it is simply the most lucid explanation of ChatGPT that has ever been written for people unschooled in the art.
TL;DR? Generative AI uses a corpus of previously-written material to generate new-ish content that is statistically derived from that corpus. In other words, the likes of ChatGPT are superb at repeating phrases that have already been uttered across all of written history, at lightning speed. And that is about it.
People are worried, as they always seem to be when it appears that the need for certain skills might disappear. Once you’ve taken it all in, however, you might feel relieved about the potential for large language models and generative AI to refine the menial work that we do so that we can focus on better things.
In the world of software engineering education, where I spend some of my most interesting off-hours, some are concerned about the ability for generative AI to interfere with learning the art of programming. Nonetheless, the best educators already have experience with the manual means to the same end: things like Stack Overflow, SourceForge, GitHub, and other similar repositories that amplify the adage that discourages us all from reinventing the wheel: “The best programmers are lazy programmers.” Because of this, these leading instructors are in the process of inverting their curricula with an emphasis on expository exercises that have students explain what their generated and third-party code is doing.
Education asks us to learn, and learning involves a balance of creation and understanding. Is one more essential than the other? Does one have to be able to create in order to understand? Or is one better off developing understanding to foster creation?
You may recall grade school science projects that involve electricity…wiring up a battery with a light bulb to make a quiz circuit; generating electricity from a potato; electromagnets; crystal radios; and so forth. My father and two of my older brothers were in the electronics industry. When I came home one afternoon in the late 1970s with my sixth grade project assignment, my family’s expectations took me by surprise. They felt I needed to present a project that plugged into a wall outlet, involving electronic components. They proceeded to conceive of a flashing neon tube project that involved a diode, a resistor, and a capacitor, similar to what you see in this video, but finished cleanly with professional soldering and clear heat-shrink tubing, installed on an attractive piece of 70s-era plywood paneling with labels on the back.
I was puzzled. Was my family encouraging me to cheat? They assured me that I wouldn’t be getting away with anything. They demanded that I learn the principles of the diode, the resistor, the capacitor, the physics behind the neon tube, and had me explain those back to them, countless times, in my own words, before I set foot in school with my assembled project.
I sat alongside them as parts were selected and as the project was assembled.
The day I walked into class with my paneling-mounted electronics, I watched a few presentations that employed D-cells and lantern batteries. When I was called, I nervously walked to the front of the room and plugged my little project into the outlet in the black-top lab desk. While I got a small thrill from being different from everyone else, I was still nervous, and I am sure I remember the teacher looking a little worried himself.
It went well. My fellow students were as astonished as I was about the bright, blinking light. We all learned something in the process. My classmates learned about things that weren’t in the curriculum, and I learned this: It’s one thing to make something; it’s a whole other thing to be able to explain how and why it works.
My teacher surprised me with an “A” grade, and I learned not only something about electronics…I learned a lesson in education that I still can’t forget.
At some point in the next 10 years, our workforce will see the demotion of scores of software engineers who eschew generative AI programming. If you don’t believe this, then ask yourself: would you, today, tolerate a software engineer or IT professional who refused to use a search engine to find solutions to a technical problem? Of course not; you’d fire them as soon as you could.
I’ve heard some software engineering instructors wonder how bad generative AI will make things for liberal arts educators. But the answers are strikingly similar on that side of campus.
In this blog, where we discuss matters relating to the nexus of liberal arts and technology, it’s worth referencing a simple but commonly-overlooked fact: writing itself is a technology. Predating the written word was the oral tradition, where people composed stories of easy-to-remember “epithets” to create stories like Homer’s Odyssey. The invention of writing liberated people from epithets, allowing people to string together create fanciful combinations of words that — to people’s horror! — could not be remembered without referring to the medium to which they were committed. If you are curious about the details of this consequential and antique technological transformation, I could not recommend a work more highly than Walter Ong’s Orality and Literacy.
Since writing is a technology — and not at all natural – we would do well to remember that enhancements to any technology are normal, and not to be considered at odds with what is natural. Much writing that we do today is what one might call “perfunctory.” Think of the vast number of forgettable emails and text messages that we hurtle back and forth each day, whose purpose is merely to drive a larger conversation about a single concept. It’s perfectly fine to have help typing those thoughts out in a way that relieves our fingers and saves us time.
We have names for certain classes of communication. Linguists have a term for the most routine communication that we employ every day: phatic. The world of generative AI presents us with an opportunity to expand our palette. Consider the following:
- Phatic communication (greetings and other similar pleasantries)
- Perfunctory communication (emails; simple essays about basic concepts; text messages; common persuasive communication; and other forgettable acts of discourse)
- High-value communication (first-person journalism; original documentary writing; poetry; creative writing; lyricism; cognitive dissonance; and other forms of inventive discourse that are designed to be memorable and durable)
Generative AI is likely to find its greatest application helping us deliver perfunctory communication with breathtaking ease and speed, in the very same way that calculators help us all with a wide variety of perfunctory mathematical tasks, allowing educators to focus on teaching skills that support high-value communication, where we ask the human mind to be entirely engaged.
Consider works such as:
- Finnegans Wake by James Joyce
- Crazy Jane Talks with the Bishop by W. B. Yeats
- The Waste Land by T. S. Eliot
- I Have A Dream by Martin Luther King
- I Am The Walrus by The Beatles
- Now/Later/Soon by Stephen Sondheim
Want to be the first person to put “Expert texpert” in front of “choking smokers?” Generative AI isn’t going to get you there. Inventive combinations of words like these are at complete odds with the statistical models behind generative AI. They are high-value in that they are landmark works that have inspired millions if not billions of people through their originality of construction. Imagine a world of liberal arts education that focuses on the ability to craft these sorts of works? The degree in “letters” might be transformed, for the better.
What does all of this portend for education in any discipline that is affected by generative AI? We would do best to ensure that we engage students to explain the reasoning behind their work in real time. This is not a new concept, but it’s an unfortunately rarified one, reserved for pivotal moments like the defense of a thesis. Education would be transformed, but teachers would have to work much harder. Of course, things that are hard are things worth doing.
Consider what it might be like to re-focus on the talents that have been neglected since the days of the oral tradition: speaking that inspires and creates movement.
Imagine a day when we frown upon PowerPoint presentations, and look forward to our fellow humans speaking extemporaneously and creatively, from their hearts, providing insight and inspiration at the times we need it most.
Imagine a day when our programmers are freed from writing login screens, and where they can focus on creating user experiences that not only save us time, but touch our hearts and souls with software that provides insight and inspiration.
Many are concerned about how “correct” generative AI is; they are alarmed by the potential effect of “hallucinations.” But these notions are not new; every book on every shelf of every library is written and edited by fallible human beings, a great deal of whom acted out of not only ignorance, but out of self-interest or with ill intent. Consumers of information have always had a duty to think critically before acting on that information. They still do.
Technology changes how we live. Writing’s initial gift was a reduction in our need to remember details. Writing’s second gift was its ability to be mass-produced, bringing us more-or-less perfect one-to-many communication. Writing’s third gift was its ability to show us how repetitive and perfunctory so much of our communication is. Generative AI gives us a chance to make perfunctory communication — and programming — even more perfunctory, liberating us for better things…if only we allow ourselves the opportunity.
Once more:
Is a written piece inherently valuable?
Does the world need more writing?
Does it need more writers?
Or would it benefit from more original thought?
Discuss this specific post on Twitter or LinkedIn.