Generative AI systems that engage in conversations and create images, songs and videos have shaken up the world. Some 300 million people use ChatGPT weekly.
The impact on software development has been even more profound. AI systems now write programs faster and more efficiently than humans. ChatGPT o3 scores higher than 99.9% of professional programmers on tests. Over 25% of Google’s recent code was AI-written. The once hot programming job market has cooled considerably.
Does this mean programming is no longer a useful skill to teach in schools? The answer requires thinking about the opportunities chatbots provide in a new way.
Value of programming in the modern world
Since 2022, I’ve been experimenting with AI tools to make computer programs. Many schools worldwide have added computer science classes.
Organizations like CS4All and Code.org promote coding as a way for young people to master the computer, possibly the most important invention of the last hundred years.
Despite this attention, opportunities remain limited. There aren’t enough teachers or time in the school day. But what if programming experiences could be easier and more integrated with other subjects? AI chatbots offer interesting opportunities to answer that question.
Using AI tools to co-create programs isn’t meant to replace teachers but to provide learner agency. In my experience, chatbots help students iterate on designs by creating something quickly, then rapidly improving it.
Critics might say creating apps quickly leads to superficial learning. But the process is conversational—you engage in design and debugging with the chatbot. Creating apps quickly allows students to learn through experience.
Making apps without learning to code
Students regularly use chatbots, too often asking them to do assignments and learning little. If instead chatbots become creative partners, students can focus on design, critical thinking, communication and problem-solving, leaving technical details to the chatbot. They can tackle far more ambitious projects.
I advocate using chatbots to co-create “web apps”—interactive browser-based applications easy to share and run safely. Browsers are already configured for student use in most schools, addressing security and accessibility concerns without additional software.
Since ChatGPT debuted, I’ve developed hundreds of examples of students co-creating apps with chatbots, from simple apps that young children can make to complex scientific and machine learning projects.
I’ve documented these methods in my book, The Learner’s Apprentice: AI and the Amplification of Human Creativity. Teaching students to guide a chatbot through app creation is an exercise in critical thinking and iterative design.
Examples of apps students can co-create include:
- A wide variety of games, from word games to graphic adventures
- Scientific explorations and simulations
- AR experiences
- Incorporating AI into apps
- Mathematical explorations from proofs to data science
- Using and training your own machine learning models
- Programming microcontrollers and mobile devices
Serving a wider range of learners
Learners can quickly co-create sophisticated apps or go slowly and learn programming languages in the process. Chatbots’ adaptability suits different learning goals.
- Chatbots provide adaptive support to students who want to:
- Design and make interactive games or tools
- Understand how apps work without learning technical details
- Learn a programming language
Chatbots open programming-like experiences to students previously uninterested in learning to program. They can make apps without spending months learning to code, while motivated students can still dive into any language.
Challenges and opportunities
Chatbots aren’t perfect. They make mistakes and hallucinate facts. Learning to fact-check chatbot responses is valuable. Chatbots sometimes produce buggy code, so students must learn the art of debugging, which develops critical computational thinking skills.
Creating apps isn’t the only creative possibility. Students can create historical text-based adventures, simulate debates and create illustrated stories on any subject. Teachers can customize chatbots for desired pedagogy, tone, and language complexity.
Powerful ideas that will never be obsolete
For 50 years, I’ve built tools to make programming accessible to non-experts, motivated by sharing the joy of creating programs and providing fertile ground for acquiring powerful ideas. I still support introducing programming to all children, but chatbots have changed what I believe the focus should be—away from writing code toward reading and understanding programs.
Some students may still want to learn programming languages, especially for software careers. But developers increasingly use AI tools to write programs. Learning to collaborate effectively with AI tools complements traditional programming education.
Students in sciences and mathematics currently learn programming because it plays a large role in STEM fields. However, this will change.
Kyle Kabasares, a data scientist, used ChatGPT o1 to replicate coding from his PhD project calculating black hole mass. “I was just in awe,” he says, noting it took o1 about an hour to accomplish what took him many months.
Most learners can more easily experience the joy of creating apps without learning to write programs. Co-creating apps with chatbots may provide even more fertile ground for acquiring advanced ideas and computational thinking, allowing more creative projects in the same timeframe.
Lower-level computational concepts remain important, but acquiring them by making apps with chatbots has many advantages over just learning a programming language.
