Revolutions are Seldom Cute
Educational rhetoric has long been rife with obfuscation, euphemism, and verbal inflation. Teacher is no longer in vogue. Facilitator or leader are preferable. Six-year-olds are called scholars. Middle managers hold titles like Assistant Director of Future. Each action is lauded as innovation. Tests are assessment. Grades are rubrics. Tracking, ranking, and sorting of students sounds kinder and gentler when renamed streaming. Art, music, dance, and drama programs are denatured as media. These are but a few examples of the affliction Seymour Papert diagnosed as verbal inflation.
One of the most overused superlatives of the AI era, is “revolutionary.” Declaring a person, product, or action as revolutionary is a very big claim — one worthy of scrutiny, or at least skepticism. Very little educational “change” truly earns the appellation of revolutionary. Count yourself blessed to encounter one true revolutionary in your lifetime, despite the chutzpah of social media and hyperbole of conference programs.
One day recently, two pieces of artificial intelligence software were released. Both are likely to be declared revolutionary technological innovations. One may be. The other certainly is not. As sure as I am making that statement, I am equally confident that the “wrong” software will get the most attention, particularly in education. Whether you believe that AI will save or kill us, opinions are extreme, polarizing, and largely inconsequential.
Sora
OpenAI released Sora, software that will generate a video from a text prompt. The results are magical, dazzling, hilariously bad, and all of the other qualities found in generative AI output. That said, I fully expect the folks who suddenly changed their LinkedIn profiles to read, “AI in Education Expert,” will gush over Sora, demonstrate it breathlessly at conferences, and cite it as revolutionary.
One of the obvious problems with Sora, aside from the typical unreliability we have come to expect from chatbots, is that the videos produced are limited to a maximum length of 15 seconds. So, it you’re interested in animated poop dancing an Irish jig while waving a Ukranian flag, you’re in business, until the algorithm decides that the flag is too political and rejects your prompt.
Will this and similar technology improve? Yes. Will the creation of longer videos be possible? Absolutely, and so what? This is certainly not revolutionary, especially in an educational context. Let’s face it, schools have short videos no one will ever watch on lock. Claymation, bitmojis, shoebox dioramas, magazine collages, coat-of-arms coloring pages, five-paragraph essays, mobiles, Ignite talks, and 15-second videos are a school’s superpower.
Language arts is the cul-de-sac edtech lives on. New tools may improve storytelling, which is an obviously good thing. However, adding another cute way to tell a trivial story hardly represents a revolution. No matter how cute Sora videos happen to be, revolutions are seldom cute.
Revolutionary Potential
My work suggests that the project should be a teacher’s smallest unit of concern. Piaget teaches us that “knowledge is a consequence of experience.” Therefore, project-based learning is an effective way of creating the conditions necessary for such knowledge constructing experiences. We have long known how to teach Language Arts, Social Studies, the arts, and even Science in a project-based fashion. This leap has been much more difficult and less obvious for mathematics teaching. We just didn’t have the materials, tools, and expertise necessary to “do Math” in project-based fashion. Computing changes that. Computational tools like the Wolfram Notebook Assistant, launched the same day as Sora, may be a game-changer.
Mathematica, the computational software environment created decades ago by Stephen Wolfram and his company, is the widely used tool by researchers engaged in mathematics, science, or engineering. It is insanely powerful and intended for people engaged in serious scientific pursuits. Mathematica’s computational power is supercharged by the giant collection of accumulated knowledge and data it can access.
Wolfram Alpha is a consumer-facing web application that does for computation what “the Google” did for search. Ask Wolfram Alpha to add 3 oranges to 2 bananas and you will get a result. That result in turn is a computational object that may be manipulated, operated upon, or interrogated further. Wolfram Alpha is not however limited to numerical calculations. It can create graphs, maps, tables, and other computational objects — or be used as a super powerful calculator. Stephen Wolfram once told me that he and his colleagues know what time school lets out everywhere on earth based on the banal “homework problems” typed into Wolfram Alpha as soon as kids get home.
Underneath Wolfram Alpha and Mathematica lies Wolfram Language. Several years ago, the Wolfram company began providing user access to that underlying language via software options, including the Wolfram Cloud allowing Wolfram Notebooks in which computational experiments, problems, and projects could be produced, annotated, modified, and shared in a word processor-like format. Now, language could be used to explain formal computations and produce transparent results open to replication, verification, and remixing by others. Complex calculations, projections, models, and dynamic simulations can often be addressed in one line of text. Making this possible via a browser meant accessibility by anyone anywhere. This represented a major step forward, but like all programming languages one needs to use proper syntax and vocabulary, regardless of how good the online Help happens to be.
On December 9, 2024, Wolfram lowered the bar of entry to this world of “computational making” by introducing Wolfram Notebook Assistant which adds the natural language capability of a chatbot, ala ChatGPT, to the Wolfram environment. This is a BFD! Now, laypeople, even children can tell the system what they would like it to do and have it produce a result without learning the system’s language, syntax rules, or idiosyncrasies.
The system can even operate on financial data, machine learning input, spreadsheets, images, video, and sound files as they too are computational objects. Ask a question or submit a request and the AI assistant will translate your natural language, warts and all, into Wolfram Language code to be executed. Not only that, but the system shows you the code and explains how it intends to go about computing a result. This not only makes the process transparent and verifiable but is invaluable for learning Wolfram Language as well. Before long, even numbskulls like me begin making changes to the code directly and forgo the need for natural language explanations.
You owe it to yourself to read Dr. Stephen Wolfram’s essay about this new technology in his essay, Useful to the Point of Being Revolutionary: Introducing Wolfram Notebook Assistant or watch the accompanying video. In the unpolished video, Wolfram engages in live-coding to demonstrate the power of this remarkable new software.
The Wolfram Notebook Assistant knows about itself to teach you how to use it and clarify your thinking. It copes with typos, a mixture of cardinal and ordinal numbers, multiple explanations, and levels of user sophistication. It can export its results in a variety of formats to be used by other tools as well. In the “Wolfram Notebook” shared below, I engaged in what I like to call generative design. The result of each step leads to a new idea or an opportunity for debugging. The computer mediates a conversation with myself while performing a remarkable quantity of computation quickly without fuss or muss.
I started with a simple prompt exploring the use of randomness to create five-letter words—easy-peasy for Wolfram Notebook Assistant. Then, I asked it to generate many of those words, and then even more. Next, I asked the system how many of them were actual English words. Yes, it knows that too. Then, I requested the percentage of random words that were real English words. Larger data sets, multiple graphical depictions, and word clouds followed.
I even asked my computational apprentice to define one of the words before investigating whether seven-letter random words behaved differently. Then, I asked for the least popular letters in English and instructed the system to generate random words without using those letters. I compared those results with my previous experiments. Of course, I could have asked Wolfram Notebook Assistant to perform that analysis as well. Along the way, it explained how to calculate percentages and even optimized its own code for efficiency. The learning potential is limitless.
Gary’s Project Notebook
I experimented with different terms, sentence structures, punctuation (and lack thereof), numerals, and written-out numbers to test the system’s flexibility. While this may not be rocket surgery, my little “project” offers a glimpse into something truly revolutionary.
Revolutions are seldom cute and often leave innocent casualties in their wake. Any educational revolution should add surplus value — leave the classroom better than we found it. Educators concerned with the future of schooling would be wise to avoid being distracted shiny objects and invest their attention in the democratization of computation. Ask yourself, “how might I achieve the maximum return-on-investment?” The Wolfram Notebook Assistant brings us much closer to Seymour Papert’s vision of Mathland, a place where mathematics is useful, relevant, beautiful, playful, and comprehensible, than ever before.
Caveat: Our friends at Wolfram have limited experience working with the K-12 school market and it is currently too expensive and confusing to make use of the Wolfram Notebook Assistant widespread in classrooms and kids’ bedrooms as it deserves to be. I have been offering my advice to colleagues at Wolfram and there may soon be a solution to this problem in a streamlined offering for primary and secondary schools. Watch this space.