About the Study

As AI assistants like Cursor, Claude, and GitHub Copilot become ubiquitous, we are entering the era of vibe coding – where the AI generates the logic while the developer guides the direction.

But what happens to our underlying skills over time? Does this convenience lead to cognitive atrophy, or does it free us to tackle higher-level problems?

Can I Still Code is a longitudinal, citizen-science research project designed to measure exactly that. Participants complete short, timed Python challenges every 28 days – no AI, no Google – and self-report how much they rely on AI tools in their day-to-day work. Over time, the resulting open dataset will allow us to model whether vibe-coding intensity predicts changes in raw coding ability.

The study runs for at least 12 months. The anonymised dataset will be published under an open licence (CC BY 4.0) after a 12-month embargo, and any resulting papers will be submitted to open-access venues.


About the Researcher

I'm Andy Woods, a researcher at Royal Holloway, University of London. My background is in experimental psychology, but I have also been writing production Python and JavaScript for about 15 years – building and deploying Django applications, and at various points working in the startup world.

My interest in developer deskilling is personal as well as scientific. I use AI coding assistants in my own work, supervise a PhD student working on AI and coding, and found myself genuinely unsure of the answer to a simple question: am I still actually any good at this?

I bring to this question the methodological tools of experimental psychology – Bayesian multilevel regression, longitudinal design – alongside the experience of a working developer who has skin in the game.

Royal Holloway profile  ·  Google Scholar  ·  LinkedIn


Ethics & Transparency

The study has received ethics approval from Royal Holloway, University of London. Participation is entirely voluntary. You can withdraw at any time and request deletion of your data.

The study is pre-registered on OSF. All analysis code will be published on GitHub. We are committed to open science at every stage.


Get Involved

Whether you are a participant, a researcher interested in collaborating, or an organisation that would like to support the work, we'd love to hear from you.