If you have spent any time around programmers, you might have noticed something odd - rubber ducks. Small, yellow ducks perched on monitors, keyboards, or desks. They aren't decorations or stress toys - they are debugging partners.
This practice, known as "rubber duck debugging," has been a programmer's trick for decades. The premise is simple: when you are stuck on a problem, explain it to the duck. In that process of articulation, one finds that they often magically discover the issue themselves.
This rubber duck strategy was popularized by Andrew Hunt and David Thomas in their 1999 book "The Pragmatic Programmer." They shared a story about a programmer who carried around a rubber duck, debugging code by explaining it line-by-line to the duck. This act of verbalization forced them to slow down and break their mind from the logic error traps one falls into in programming. Ultimately, it greatly speeds up the debugging time and allows for faster development.
Programmers have long known that explaining a problem to someone else often leads to breakthroughs. The rubber duck formalized this insight and removed the need for a colleague pretending to listen.
Fast forward to today, and we have something that was mere science fiction in 1999: AI assistants that can actually understand what you are saying and respond intelligently. However, many don't know how to use it properly. Many times, a programmer will simply say: "do this task", "make this app", or "fix this bug." Sometimes this method works, often it doesn't. But for a professional programmer, the best way to use your AI agent is an upgraded form of the method they were already using - the rubber duck.
By treating your AI assistant (be it Chat-GPT, Gemini, Claude, etc.) as a Rubber Duck with a Brain, you get a mediary to bounce your ideas off of, a mirror to reflect your own logic in, a compiler for your mental processes. You can ask AI to write some code, sure, but have you ever tried to ask it for thoughts on your ideas and concepts - BEFORE writing any code? I've found that this method supercharges my development - suddenly, I'm not just asking for AI to create an arbitrary thing, I'm asking it to execute on a thorough development plan that I came up with by bouncing my ideas back and forth, clarfiying them in my own head.
Is it pure automation? No. But you will find that effectively treating your AI assistant as that magical "Rubber Duck with a Brain", you will be able to produce better results faster than simply trying to create and recreate arbitrary ideas that neither you nor the AI are able to firmly nail down. So next time you sit down with a coding project, try asking your helpful AI assistant "what do you think of my plan" a few times before "build my plan."