AI is Like Steroids for Programmers and Should Be Illegal
A programmer just shipped a feature in 2 hours that would have taken 2 weeks. Suspicious? Absolutely.
Just as baseball had its steroid scandal in the early 2000s, programming is facing its own performance-enhancing drug crisis. Barry Bonds hit 73 home runs in 2001. Mark McGwire and Sammy Sosa shattered records. And now? Programmers are shattering velocity records with AI tools.
The question we must ask: Should we ban AI tools to preserve the “natural” art of programming?
The Parallels Are Undeniable
Consider the evidence:
Unfair advantages. Barry Bonds hit 73 home runs in a single season. You’re hitting 73 commits per day. Coincidence? I think not.
Unnatural enhancement. Anabolic steroids build muscle mass at rates that exceed natural human capability. AI generates code at rates that exceed natural typing speed. Both bypass the fundamental limits of human biology.
Elite performance. Olympic athletes get disqualified and stripped of medals for doping. Should 10x engineers get disqualified for prompting? Where’s the integrity?
The asterisk. Baseball fans still debate whether to put an asterisk next to steroid-era records. “Sure, you built that feature in a weekend… with Claude.” Same energy.
The Case for AI Testing
If we’re serious about protecting the integrity of our craft, we need to implement strict testing protocols:
Random AI screenings during standup. Your velocity doubled this sprint? Time for a surprise code review. No AI-generated patterns allowed.
Separate leaderboards. We need clean programming records vs. AI-assisted records. Linus Torvalds built Linux without Copilot. His achievements count. Yours? Enhanced.
Hall of Fame debates. In 50 years, will we look back and say, “The 2020s era doesn’t count because everyone was using AI”? We need to preserve the pre-AI records as sacred.
The purists are right. Real programmers use Vim and StackOverflow, the way God intended. Not these new-fangled LLM crutches.
Gateway drug concerns. First it’s autocomplete. Then it’s Copilot. Then it’s full AI agents writing entire features. Where does it end? When the AI is interviewing you for permission to commit to your own repository?
Following the Logic
If AI is cheating, then we must be consistent.
IDEs are cheating. Syntax highlighting gives an unfair advantage over notepad. Ban it.
Compilers are cheating. Real programmers write assembly. High-level languages are performance-enhancing compared to machine code.
High-level languages are cheating. C is basically steroids for programmers compared to punch cards. We should all go back to physically rewiring computers like the ENIAC operators.
Stack Overflow is cheating. Looking up answers? That’s not your knowledge. That’s crowd-sourced doping.
Books are cheating. Reading K&R to learn C is basically ingesting performance-enhancing knowledge. Real programmers figure everything out from first principles.
Electricity is cheating. Alan Turing did computations by hand. We should too.
Wait.
The Actual Point
Every generation of tools was called “cheating” by the previous one.
Assembly programmers said C was cheating. C programmers said Python was cheating. Vim users said IDEs were cheating. And now, text editor purists say AI is cheating.
The pattern is obvious: tools that amplify human capability always face resistance from people who mastered the previous tools.
But here’s the difference between steroids and AI:
Steroids bypass biology. You inject them and your body builds muscle without proportional effort. The shortcut is the entire mechanism.
AI doesn’t bypass thinking. You still need to:
- Understand the problem
- Architect the solution
- Review the generated code
- Debug when it breaks
- Maintain it over time
AI is a tool. A very powerful tool. But it doesn’t do your thinking for you. It amplifies your thinking, the same way a compiler amplifies your assembly code, or a framework amplifies your boilerplate.
As I wrote in my project guidelines: “Quality-conscious: ‘no AI slop’—reviews all AI-generated code.” That’s the key. The tool doesn’t make you lazy unless you let it.
The Real Danger
The actual risk isn’t that AI makes us lazy. It’s that we forget to verify its work.
Just like steroids without training creates bulk without functional strength, AI without understanding creates code without maintainability.
The scandal isn’t using AI. The scandal is shipping garbage and blaming the AI.
The scandal is not reading what Claude wrote before you commit it.
The scandal is treating AI like magic instead of like a very fast, very confident junior developer who sometimes hallucinates.
So Yes, I Use AI
And yes, I review every line.
And yes, I understand what it’s doing.
And yes, I architect the systems it implements.
Call it cheating if you want. I call it 2025.
The question isn’t whether to use AI. The question is whether you’re disciplined enough to use it well. Whether you have the judgment to know when it’s right and when it’s hallucinating. Whether you can review code faster than most people can write it.
That’s the new skill. That’s what separates the enhanced programmers from the sloppy ones.
Barry Bonds may have used steroids, but he still had to swing the bat. And he still had to hit a 95 mph fastball.
I use AI. But I still have to think.
This blog post was written with AI assistance. Judge for yourself.