Today we’re publishing a position paper setting out five arguments for why we think that kids still need to learn to code in the age of artificial intelligence.

Just like every wave of technological innovation that has come before, the advances in artificial intelligence (AI) are raising profound questions about the future of human work. History teaches us that technology has the potential to both automate and augment human effort, destroying some jobs and creating new ones. The only thing we know for sure is that it is impossible to predict the precise nature and pace of the changes that are coming. 

One of the fastest-moving applications of generative AI technologies are the systems that can generate code. What started as the coding equivalent of autocomplete has quickly progressed to tools that can generate increasingly complex code from natural language prompts. 

This has given birth to the notion of “vibe-coding” and led some commentators to predict the end of the software development industry as we know it. It shouldn’t be a surprise then that there is a vigorous debate about whether kids still need to learn to code. 

In the position paper we put forward five arguments for why we think the answer is an unequivocal yes.

We need humans who are skilled programmers 

First, we argue that even in a world where AI can generate code, we need skilled human programmers who can think critically, solve problems, and make ethical decisions. The large language models that underpin these tools are probabilistic systems designed to provide statistically acceptable outputs and, as any skilled software engineer will tell you, simply writing more code faster isn’t necessarily a good thing. 

Learning to code is an essential part of learning to program

Learning to code is the most effective way we know for a young person to develop the mental models and fluency to become a skilled human programmer. The hard cognitive work of reading, modifying, writing, explaining, and testing code is precisely how young people develop a deep understanding of programming and computational thinking. 

Learning to code will open up even more opportunities in the age of AI 

While there’s no doubt that AI is going to reshape the labour market, the evidence from history suggests that it will increase the reach of programming and computational approaches across the economy and into new domains, creating demand for humans who are skilled programmers. We also argue that coding is no longer just for software engineers, it’s becoming a core skill that enables people to work effectively and think critically in a world shaped by intelligent machines. From healthcare to agriculture, we are already seeing demand for people who can combine programming with domain-specific skills and craft knowledge. 

Coding is a literacy that helps young people have agency in a digital world

Alongside the arguments for coding as a route to opening up economic opportunities, we argue that coding and programming gives young people a way to express themselves, to learn, and to make sense of the world. 

And perhaps most importantly, that learning to code is about power. Providing young people with a solid grounding in computational literacy, developed through coding, helps ensure that they have agency. Without it, they risk being manipulated by systems they don’t understand. As Rushkoff said: “Program, or be programmed”.  

The kids who learn to code will shape the future

Finally, we argue that the power to create with technology is already concentrated in too small and homogenous a group of people. We need to open up the opportunity to learn to code to all young people because it will help us mobilise the full potential of human talent, will lead to more inclusive and effective digital solutions to the big global challenges we face, and will help ensure that everyone can share in the societal and economic benefits of technological progress. 

The work we need to do 

We end the paper with a call to action for all of us working in education. We need to challenge the false narrative that AI is removing the need for kids to learn to code, and redouble our efforts to ensure that all young people are equipped to take advantage of the opportunities in a world where AI is ubiquitous.

You can read the full paper here:

---

The cartoon image for this blog was created using ChatGPT-4o, which was prompted to produce a “whimsical cartoon that expresses some of the key ideas in the position paper”. It took several iterations.