Democratizing the future of education

We are all EdTech designers, now

I have spent over a decade working on how to use technology to teach business skills at scale. I have focused on this specific issue because we have learned from multiple controlled studies that even small amounts of business education changes lives. Some examples: Ugandan founders randomized into a three week mini-MBA startup course had ~25% higher profits three years later; teaching business social skills in Tonga boosted profits of founders by 20%; founders in Singapore who learn basic business skills grow their businesses twice as fast. These are profound impacts, but they require getting education to people who need it, while delivering it to them in a way that can be helpful and impactful. And, obviously, the benefits of learning go beyond business education: "education appears to be the most consistent, robust and durable method yet to be identified for raising intelligence."But how can we deliver it to everyone, tailored to their level and ability?

The solution has always seemed to be through using education technology (EdTech to its friends). But every EdTech solution has fallen short of the dream of providing high-end education, as we've discovered the limitations of various programs that ranged from providing kids with free laptops to creating massive video courses. With a team of amazing people, I started Wharton Interactive, which uses games and simulations to teach business lessons, to try to improve on these approaches. While our games work well (you can play free games about teaching and entrepreneurship if you want to try them), it is challenging to make as many games as the world needs, as quickly as it needs them. Other ambitious EdTech projects have also run into similar issues deploying high-quality products at scale. Progress is being made, but it is not fast enough.

But now, something has changed. As of the past month, the teachers of billions of people around the world have access to a tool that can potentially act as the ultimate education technology. This tool that can create lessons that take advantage of the latest in pedagogical science, while doing so in a way that matches both the students and the local context. I am, of course, talking about AI, and specifically GPT-4. While OpenAI charges for access, it is also available, for free through Bing’s Creative Mode, to people in 169 countries. And I think this changes everything for education.

Every Teacher as EdTech Designer

Why does this matter? Because, with just a paragraph of instructions, teachers can create tools to help their students learn that would have, in previous years, required buying specialized software built by EdTech companies, even assuming that those companies existed. As of now, all you need is a prompt and some time to experiment.

This is not an exaggeration, what you can produce from a single prompt is extraordinary. In one of our papers, we outline the details behind five different prompts teachers can use, but take, for example, this simple instruction as a starting point:

You generate clear, accurate examples for students of concepts. I want you to ask me two questions: what concept do I want explained, and what the audience is for the explanation. Provide a clear, multiple paragraph explanation of the concept using specific example and give me five analogies I can use to understand the concept in different ways.

Paste it into Bing creative mode (really, you should try it now), and you will see a few things:

1. Bing AI will execute the instructions, asking you for the concept and what you want to teach. (I randomly chose for our example to reach out to middle schoolers in the US about the concept of entropy)

2. When you answer, it will do a web search on the topic, which reduces the chance that the AI will come up with incorrect answers (though you always need to double-check carefully, as the AI can make up answers that are wrong, but seem convincing). Bing provides sources for its information, which can make fact-checking easier.

3. The AI will then provide you with the explanation you needed. Since there is some randomness in the system, it will provide slightly different answers each time, so you can always try again if you don’t like the results. Or, even better, ask the AI to change something: “can you swap out the third example for another” or “can you make the explanations simpler”

But we can go further than even these good explanations. How about adding some low-stakes testing: You are a quiz creator of highly diagnostic quizzes. You will make good low-stakes tests and diagnostics. You will construct several multiple choice questions to quiz my audience on the topic. The questions should be highly relevant and go beyond just facts. Multiple choice questions should include plausible, competitive alternate responses and should not include an "all of the above option." At the end of the quiz, you will provide an answer key and explain the right answer.

Now we get a quiz that we can use for students (again, we want to check the answers carefully)

This still seems a little boring. Make a in-class activity for middle schoolers to learn about entropy. Make sure it involves them moving around. Give me the exact instructions to give the class and the way to run the exercise. Impressvely, it produces a pretty interesting class exercise (this is the first time I ever tried this prompt, by the way, it just did this without any edits or feedback)

Or even further… give me code for a small game that teaches a concept related to entropy for middle schoolers.

🤯It creates working Python code for a game where I can increase or decrease entropy in a system by pushing the error keys. And the instructions for using it. Again, this is on the first try, I never tried this prompt before.

Below is a gif of the working game, for which I basically just pasted the text into a notebook as the AI instructed and it ran (there was one small error, I told the AI, and it revised the code and fixed the problem). If I hit up arrow, it increases the entropy of the system, and the molecules move faster, knocking each other apart. I hit down arrow, and the system moves slower, allowing molecules to reform. Crude, but what a start from a sentence-long prompt.

And this is just the start. Experimentation will show that you can create simulations, learning games, grading rubrics, assignments, and much more.

Democratizing EdTech

Once an exclusive privilege of million-dollar budgets and expert teams, EdTech now rests in the hands of educators. While it's important to remain vigilant for potential hallucinations, errors, and biases, GPT-4 enables teachers to craft personalized prompts tailored to their local contexts, significantly bolstering their resources in the pursuit of quality education.

It's time for governments and philanthropic organizations to contemplate their roles in facilitating the development and dissemination of these novel EdTech solutions, while also guiding educators towards measurable, impactful changes using these methods. EdTech providers must also adapt to the shifting landscape and actively participate in democratizing access to these cutting-edge tools by broadening their capabilities.

We find ourselves in an exhilarating era for educators: the power to devise customized solutions for our students lies in our hands, rather than relying solely on external expertise. The key to success lies in fostering a culture of ethical experimentation and collaboration, keeping the well-being of our students, schools, and societies at the forefront of our efforts.

Feel free to share some prompts and ideas that work for you in the comments.