Vibe coding outsources judgement along with execution
The term describes using AI tools without accountability, accepting whatever is produced without scrutiny. Which tools you use, and how often, has nothing to do with it. What matters is whether you keep your understanding and direction throughout, or hand off the thinking along with the work.
Vibe coding
One-sentence definition: Vibe coding is working with AI in a way that outsources judgement along with execution, using the tools while giving up accountability for the outputs.
Andrej Karpathy (2025) coined the term for a way of programming in which you “fully give in to the vibes” and stop reading the code the AI writes. Simon Willison (2025a) pointed out that most AI-assisted programming isn’t that, and later developed the idea by distinguishing vibe coding from vibe engineering: what experienced practitioners do when they use the same AI tools while remaining accountable for every output (Willison, 2025b). The tools are the same in both cases, and what differs is what the practitioner keeps hold of throughout. I use the term here in that broader sense, for any work handed to AI in this way, whether or not it involves code.
What disciplined AI use keeps
Disciplined AI use means keeping the ability to evaluate what an agent produces: recognising when something is wrong and not merely plausible, and redirecting it precisely enough to improve it. Vibe coding can produce fluent, convincing output, but you can’t stand behind that output, because standing behind it needs the judgement that was given up in producing it.
The failure shows most clearly when something goes wrong. A practitioner who has been vibe coding can’t say why the output falls short or how to improve it, so they’re left to accept or reject it without being able to direct the next attempt. In practice it looks like:
- accepting a generated literature review without checking whether the sources say what the AI claims they say
- delegating a task without a clear brief and accepting whatever comes back
- producing writing that sounds like your thinking but doesn’t reflect it
Vibe coding in health professions education
Educators have started using AI to build their own teaching tools, and some write about vibe coding approvingly, as the route by which a clinical teacher with no programming background can build a simulation or a quiz (Chow & Ng, 2025; O’Connor et al., 2026). The opportunity is real. So is the risk, and it lies in exactly the step vibe coding skips: a drug-calculation practice app for nursing students that nobody has checked will teach its errors to every student who uses it. Building the tool with AI is fine; releasing it to students without checking what it does is vibe coding in the sense this note means.
Where the line sits for substantive work
The line between disciplined delegation and vibe coding isn’t always sharp. For routine tasks — reformatting, reorganising, converting between structures — there may be little difference between accepting output and confirming it’s correct. The harder question is where the line sits for intellectually substantive work, and whether it moves as models improve and practitioners get better tools for evaluating what agents produce.
Sources
- Chow, M., & Ng, O. (2025). From technology adopters to creators: Leveraging AI-assisted vibe coding to transform clinical teaching and learning. Medical Teacher, 47(12), 1927–1929. https://doi.org/10.1080/0142159X.2025.2488353
- Karpathy, A. [@karpathy]. (2025, February 2). There’s a new kind of coding I call “vibe coding” [Post]. X. https://x.com/karpathy/status/1886192184808149383
- O’Connor, S., Zhang, M., Hui, V., et al. (2026). The era of agentic AI: Should nurses and midwives learn about vibe coding and AI agents for better health and care? Nurse Education in Practice, 93, 104837. https://doi.org/10.1016/j.nepr.2026.104837
- Willison, S. (2025a, March 19). Not all AI-assisted programming is vibe coding (but vibe coding rocks). Simon Willison’s Weblog. https://simonwillison.net/2025/Mar/19/vibe-coding/
- Willison, S. (2025b, October 7). Vibe engineering. Simon Willison’s Weblog. https://simonwillison.net/2025/Oct/7/vibe-engineering/
Notes
The contrast with context engineering is instructive: context engineering is the practice that makes disciplined AI use possible at scale, and vibe coding is what happens in its absence.