For the previous decade and a half, I’ve been exploring the intersection of expertise, training, and design as a professor of cognitive science and design at UC San Diego. A few of you may need learn my latest piece for O’Reilly Radar the place I detailed my journey including AI chat capabilities to Python Tutor, the free visualization software that’s helped hundreds of thousands of programming college students perceive how code executes. That have acquired me fascinated with my evolving relationship with generative AI as each a software and a collaborator.
I’ve been intrigued by this rising apply referred to as “vibe coding,” a time period coined by Andrej Karpathy that’s been making waves in tech circles. Simon Willison describes it completely: “Once I discuss vibe coding I imply constructing software program with an LLM with out reviewing the code it writes.” The idea is each liberating and barely terrifying—you describe what you want, the AI generates the code, and also you merely run it with out scrutinizing every line, trusting the general “vibe” of what’s been created.
My relationship with this method has developed significantly. In my early days of utilizing AI coding assistants, I used to be that one who meticulously reviewed each single line, typically rewriting important parts. However as these instruments have improved, I’ve discovered myself step by step letting go of the steering wheel in sure contexts. But I couldn’t totally embrace the pure “vibe coding” philosophy; the professor in me wanted some high quality assurance. This led me to develop what I’ve come to name “vibe checks”—strategic verification factors that present confidence with out reverting to line-by-line code evaluations. It’s a center path that’s labored surprisingly properly for my private initiatives, and as we speak I wish to share some insights from that journey.
Vibe Coding in Follow: Changing 250 HTML Information to Markdown
I’ve discovered myself more and more turning to vibe coding for these one-off scripts that resolve particular issues in my workflow. These are sometimes duties the place explaining my intent is definitely simpler than writing the code myself, particularly for knowledge processing or file manipulation jobs the place I can simply confirm the outcomes.
Let me stroll you thru a latest instance that completely illustrates this method. For a category I train, I had college students submit responses to a survey utilizing a proprietary internet app that supplied an HTML export possibility. This left me with 250 HTML information containing precious scholar suggestions, however it was buried in a multitude of pointless markup and styling code. What I actually needed was clear Markdown variations that preserved simply the textual content content material, part headers, and—critically—any hyperlinks college students had included of their responses.
Quite than penning this conversion script myself, I turned to Claude with a simple request: “Write me a Python script that converts these HTML information to Markdown, preserving textual content, primary formatting, and hyperlinks.” Claude recommended utilizing the BeautifulSoup library (a strong selection) and generated an entire script that will course of all information in a listing, making a corresponding Markdown file for every HTML supply.
(On reflection, I spotted I in all probability might have used Pandoc for this conversion process. However within the spirit of vibe coding, I simply went with Claude’s suggestion with out overthinking it. A part of the attraction of vibe coding is bypassing that analysis part the place you evaluate completely different approaches—you simply describe what you need and roll with what you get.)
True to the vibe coding philosophy, I didn’t evaluate the generated code line by line. I merely saved it as a Python file, ran it on my listing of 250 HTML information, and waited to see what occurred. This “run and see” method is what makes vibe coding each liberating and barely nerve-wracking—you’re trusting the AI’s interpretation of your wants with out verifying the implementation particulars.
Belief and Threat in Vibe Coding: Working Unreviewed Code
The second I hit “run” on that vibe-coded script, I spotted one thing which may make many builders cringe: I used to be executing utterly unreviewed code on my precise pc with actual knowledge. In conventional software program growth, this may be thought of reckless at finest. However the dynamics of belief really feel completely different with trendy AI instruments like Claude 3.7 Sonnet, which has constructed up a repute for producing moderately protected and useful code.
My rationalization was partly primarily based on the script’s restricted scope. It was simply studying HTML information and creating new Markdown information alongside them—not deleting, modifying present information, or sending knowledge over the community. After all, that’s assuming the code did precisely what I requested and nothing extra! I had no ensures that it didn’t embrace some sudden habits since I hadn’t checked out a single line.
This highlights a belief relationship that’s evolving between builders and AI coding instruments. I’m far more keen to vibe code with Claude or ChatGPT than I’d be with an unknown AI software from some obscure web site. These established instruments have reputations to take care of, and their father or mother firms have robust incentives to stop their methods from producing malicious code.
That mentioned, I’d like to see working methods develop a “restricted execution mode” particularly designed for vibe coding situations. Think about having the ability to specify: “Run this Python script, however solely permit it to CREATE new information on this particular listing, stop it from overwriting present information, and block web entry.” This light-weight sandboxing would offer peace of thoughts with out sacrificing comfort. (I point out solely proscribing writes moderately than reads as a result of Python scripts sometimes have to learn numerous system information from throughout the filesystem, making learn restrictions impractical.)
Why not simply use VMs, containers, or cloud companies? As a result of for personal-scale initiatives, the comfort of working instantly alone machine is tough to beat. Organising Docker or importing 250 HTML information to some cloud service introduces friction that defeats the aim of fast, handy vibe coding. What I need is to take care of that comfort whereas including simply sufficient security guardrails.
Vibe Checks: Easy Scripts to Confirm AI-Generated Code
OK now come the “vibe checks.” As I discussed earlier, the great factor about these private knowledge processing duties is that I can typically get a way of whether or not the script did what I meant simply by inspecting the output. For my HTML-to-Markdown conversion, I might open up a number of of the ensuing Markdown information and see in the event that they contained the survey responses I anticipated. This handbook spot-checking works moderately properly for 250 information, however what about 2,500 or 25,000? At that scale, I’d want one thing extra systematic.
That is the place vibe checks come into play. A vibe examine is basically a less complicated script that verifies a primary property of the output out of your vibe-coded script. The important thing right here is that it needs to be a lot less complicated than the unique process, making it simpler to confirm its correctness.
For my HTML-to-Markdown conversion challenge, I spotted I might use a simple precept: Markdown information needs to be smaller than their HTML counterparts since we’re stripping away all of the tags. But when a Markdown file is dramatically smaller—say, lower than 40% of the unique HTML measurement—which may point out incomplete processing or content material loss.
So I went again to Claude and vibe coded a examine script. This script merely:
- Discovered all corresponding HTML/Markdown file pairs
- Calculated the scale ratio for every pair
- Flagged any Markdown file smaller than 40% of its HTML supply
And lo and behold, the vibe examine caught a number of information the place the conversion was incomplete! The unique script had did not correctly extract content material from sure HTML buildings. I took these problematic information, went again to Claude, and had it refine the unique conversion script to deal with these edge instances.
After a couple of iterations of this suggestions loop—convert, examine, determine points, refine—I finally reached a degree the place there have been no extra suspiciously small Markdown information (properly, there have been nonetheless a couple of beneath 40%, however handbook inspection confirmed these have been appropriate conversions of HTML information with unusually excessive markup-to-content ratios).
Now you would possibly moderately ask: “When you’re vibe coding the vibe examine script too, how have you learnt that script is appropriate?” Would you want a vibe examine on your vibe examine? After which a vibe examine for that examine? Nicely, fortunately, this recursive nightmare has a sensible resolution. The vibe examine script is often an order of magnitude less complicated than the unique process—in my case, simply evaluating file sizes moderately than parsing complicated HTML. This simplicity made it possible for me to manually evaluate and confirm the vibe examine code, even whereas avoiding reviewing the extra complicated unique script.
After all, my file measurement ratio examine isn’t excellent. It may’t inform me if the content material was transformed with the correct formatting or if all hyperlinks have been preserved appropriately. But it surely gave me an inexpensive confidence that no main content material was lacking, which was my main concern.
Vibe Coding + Vibe Checking: A Pragmatic Center Floor
The take-home message right here is straightforward however highly effective: Once you’re vibe coding, at all times construct in vibe checks. Ask your self: “What less complicated script might confirm the correctness of my most important vibe-coded resolution?” Even an imperfect verification mechanism dramatically will increase your confidence in outcomes from code you by no means truly reviewed.
This method strikes a pleasant stability between the velocity and inventive move of pure vibe coding and the reliability of extra rigorous software program growth methodologies. Consider vibe checks as light-weight exams—not the excellent take a look at suites you’d write for manufacturing code, however sufficient verification to catch apparent failures with out disrupting your momentum.
What excites me concerning the future is the potential for AI coding instruments to recommend applicable vibe checks mechanically. Think about if Claude or comparable instruments couldn’t solely generate your requested script but in addition proactively provide: “Right here’s a easy verification script you would possibly wish to run afterward to make sure every thing labored as anticipated.” I think if I had particularly requested for this, Claude might have recommended the file measurement comparability examine, however having this constructed into the system’s default habits could be extremely precious. I can envision specialised AI coding assistants that function in a semi-autonomous mode—writing code, producing applicable checks, working these checks, and involving you solely when human verification is really wanted.
Mix this with the type of sandboxed execution atmosphere I discussed earlier, and also you’d have a vibe coding expertise that’s each releasing and reliable—highly effective sufficient for actual work however with guardrails that stop catastrophic errors.
And now for the meta twist: This whole weblog submit was itself the product of “vibe running a blog.” At first of our collaboration, I uploaded my earlier O’Reilly article,”Utilizing Generative AI to Construct Generative AI” as a reference doc. This gave Claude the chance to research my writing model, tone, and typical construction—very similar to how a human collaborator would possibly learn my earlier work earlier than serving to me write one thing new.
As a substitute of writing your complete submit in a single go, I broke it down into sections and supplied Claude with an overview for every part one after the other. For each part, I included key factors I needed to cowl and generally particular phrasings or ideas to incorporate. Claude then expanded these outlines into totally shaped sections written in my voice. After every part was drafted, I reviewed it—my very own model of a “vibe examine”—offering suggestions and requesting revisions till it matched what I needed to say and the way I needed to say it.
This iterative, section-by-section method mirrors the vibe coding methodology I’ve mentioned all through this submit. I didn’t want to put in writing each sentence myself, however I maintained management over the course, messaging, and closing approval. The AI dealt with the execution particulars primarily based on my high-level steering, and I carried out verification checks at strategic factors moderately than micromanaging each phrase.
What’s significantly fascinating is how this course of demonstrates the identical ideas of belief, verification, and iteration that I advocated for in vibe coding. I trusted Claude to generate content material in my model primarily based on my outlines, however I verified every part earlier than shifting to the following. When one thing didn’t fairly match my intent or tone, we iterated till it did. This balanced method—leveraging AI capabilities whereas sustaining human oversight—appears to be the candy spot for collaborative creation, whether or not you’re producing code or content material.
Epilogue: Behind the Scenes with Claude
[Claude speaking]
Trying again at our vibe running a blog experiment, I ought to acknowledge that Philip famous the ultimate product doesn’t totally seize his genuine voice, regardless of having his O’Reilly article as a reference. However in step with the vibe philosophy itself, he selected to not make investments extreme time in countless refinements—accepting good-enough moderately than excellent.
Working section-by-section with out seeing the complete construction upfront created challenges, just like portray elements of a mural with out seeing the whole design. I initially fell into the lure of copying his define verbatim moderately than remodeling it correctly.
This collaboration highlights each the utility and limitations of AI-assisted content material creation. I can approximate writing kinds and develop outlines however nonetheless lack the lived expertise that provides human writing its genuine voice. The very best outcomes got here when Philip supplied clear course and suggestions.
The meta-example completely illustrates the core thesis: Generative AI works finest when paired with human steering, discovering the precise stability between automation and oversight. “Vibe running a blog” has worth for drafts and descriptions, however like “vibe coding,” some type of human verification stays important to make sure the ultimate product actually represents what you wish to say.
[Philip speaking so that humans get the final word…for now]
OK, that is the one half that I wrote by hand: My parting thought when studying over this submit is that I’m not pleased with the writing high quality (sorry Claude!), but when it weren’t for an AI software like Claude, I’d not have written it within the first place as a result of lack of time and vitality. I had sufficient vitality as we speak to stipulate some tough concepts, then let Claude do the “vibe running a blog” for me, however not sufficient to completely write, edit, and fret over the wording of a full 2,500-word weblog submit all on my own. Thus, similar to with vibe coding, one of many nice joys of “vibe-ing” is that it significantly lowers the activation vitality of getting began on inventive personal-scale prototypes and tinkering-style initiatives. To me, that’s fairly inspiring.
