Sean Grove from OpenAI gave a talk at the AI Engineer World's Fair. He talked about a time when being clear, not just coding, will help us build with AI.
I saw Sean Grove's talk from OpenAI, and it got me thinking differently about being a coder when AI is all around. He talked about "The New Code," where the key skill isn't just making perfect code. It's about being clear in what we need and want. This, he said, is where we are going.
More Than Just Code
Grove started with a question: "Is your job to write code?" Many coders raised their hands. But, he said that actual coding is just 10-20% of what we do. The rest, 80-90%, is "structured communication." This means all the talks and plans we make to decide what to build. This hit me hard. As AI gets better at coding, our real job is to set the problem and the goal right.
Why Your Prompts Matter a Lot
We all do what Grove calls "vibe-coding." You get a feeling, turn it into a prompt for an AI, and it gives back some code. He says these prompts are like a new source code. But we throw them away once we get our code. Grove says it's like tossing your source code but keeping the final program. The real worth - the reason and the logic behind it - is lost.
A Specification led Future
This isn't just a dream. OpenAI uses this method with their "Model Spec," an open document that lists what their AI models should do and value. It's a live document that gets everyone on the same page, from coders to legal teams. They use it as a "trust anchor." If a model acts weird, they look at the spec and say, "That's a bug," because it goes against their plan.
He said a good spec keeps everything intact. You can use it for lots of things: code in different programming languages, API guides, tutorials, and even blog posts. The spec is the truth that ties everything else.
The Big Idea
What does this mean for us? Our jobs are changing. The new rare skill is crafting specs that catch the full purpose and values. In this new age, the best communicator - whether a project manager, a lawyer, or a coder - is the most valuable coder.
This is an exciting change. It shows coding was never just about the code; it was about using exact words to fix human problems. Now, that language is turning more human.