Is Computer Programming Dead

Should we still teach programming in a world where AI can write code in minutes? Short Answer: Absolutely yes, though we need to evolve our approach dramatically.

This weekend, I experimented with several AI Agents for code generation. Cursor, an AI coding agent caught my eye in Particular. The results were truly impressive - in under five minutes, I built an application that allows users to upload grayscale images, colorise them using advanced methods, detect faces, and even pushed the code to GitHub with a README file, all without writing a single line of code myself.

This experience highlighted an important truth: traditional programming education needs to be evolved, if not completely transformed. Just as we moved away from teaching assembly language as computing evolved, we must now adapt to the AI-assisted programming paradigm, and help our students to be prepared for a very competitive working environment.

I don't have any answer on how we can do this, and where to start? However, I believe that our focus should shift from teaching and heavily focusing on the basic syntax and implementation details to:

• Understanding system architecture and design within the context of the AI landscape
• Evaluating and using AI agents to generate code for complex problems, not just simple tasks (e.g., beyond sorting arrays or writing basic SQL database updates)
• Learning to effectively prompt and guide AI tools to accomplish specific objectives
• Introducing computing students to machine learning and deep learning concepts from their first year, even at a high level
• Creating awareness about the vast ecosystem of pre-trained and open-source models that can be leveraged
• Placing greater emphasis on helping students identify problems, articulate and justify them, and ultimately employ AI agents as solution partners

Students should understand how to identify and utilize existing ML/DL models rather than building everything from scratch. There's an incredible wealth of open-source models that can be integrated into applications to solve complex problems like computer vision, natural language processing, and predictive analytics. Teaching students how to effectively tap into these resources will enable them to build sophisticated applications efficiently.

Programming education isn't becoming irrelevant - I think, it's becoming more important than ever, but in a different form. The programmers of tomorrow won't be measured by how many lines of code they write, or the syntax they remember, but by how effectively they can leverage AI tools and pre-trained models to solve complex problems.

Here is the github repo, all generated using the Cursor IDE https://github.com/heyad/CvDemo