Rising Tensions in the AI Code Wars

This is The Stepback, a weekly newsletter dissecting a crucial story from the tech world. For more on the AI coding and vibe-coding booms, follow David Pierce. The Stepback lands in our subscribers’ inboxes at 8AM ET. Sign up for The Stepback here.

Writing code was a game-changer for AI even before the concept of AI gained mainstream attention. In the spring of 2021, long before “ChatGPT” became a household name, Microsoft introduced a collaborative project with a nonprofit organization called OpenAI. This initiative resulted in the creation of GitHub Copilot, a tool designed to observe developers as they wrote code and offer autocomplete suggestions. Despite being in a “restricted technical preview” phase and not yet fully refined, over a million developers eagerly signed up to test it.

The potential for large language models to streamline software development was evident. Coding languages are typically well-documented, and a vast repository of code exists online for training models. Companies envisioned that these models could not only accelerate code writing by predicting the next word, akin to Google’s autocomplete feature, but eventually automate significant portions of the coding process. This foresight led to the birth of AI coding tools.