Code was our medium for thought (wattenberger.com)

A new discourse around AI in software development posits that models and tools should not merely expedite coding but enhance collaborative problem-solving and project exploration. As Epoch AI highlights, the journey from writing code to producing a finished product has evolved through various computing eras. Today, many developers find themselves overwhelmed by a backlog of pull requests and increasingly reliant on AI agents to produce code rapidly, leading to a decline in meaningful engagement and team dynamics. Instead of using AI solely to generate code, the focus is shifting towards how humans can best interact with their ideas and the coding process. The significance of this shift lies in the potential of AI to serve as an adaptable medium for thought, enhancing the way teams collaborate and innovate. The proposal includes envisioning custom development environments where code is not the primary output but a tool for exploration—utilizing diagrams, prototypes, and other visual aids. This new paradigm has the promise of restoring intuitive understanding and aligning team efforts, emphasizing a future where coding goes hand-in-hand with creativity and exploration, challenging the notion that efficiency must come at the expense of engagement and clarity.