Learn Programming Again

A clever nod to Apple TV+'s "Severance" where the character's work self ("innie") is separated from their outside self ("outie"). Just like how developers have two personalities: the one who stubbornly refuses AI assistance at home, and the one who secretly lets GitHub Copilot write half their codebase at work. We all have principles until the deadline is tomorrow.

When your neural network confidently labels a cat as a dog, but everyone's freaking out about the AI apocalypse. Look, I've been training models for 15 years, and I can assure you the biggest threat isn't Skynet—it's that production code written at 3 AM with no code review. The real uprising will start when my model can correctly identify my cat and remember to order cat food when I'm running low. Until then, we're safe from the robot overlords... probably.

Every developer's relationship with Copilot in two frames. First you're all polite about it, nodding along like "ah yes, very innovative, love what you've done with the place." Then reality kicks in and you're frantically googling how to turn off the AI that keeps autocompleting your variable names into Shakespearean sonnets. It's like having an overly enthusiastic intern who won't stop suggesting "improvements" to your perfectly functional code. Sure, it can write a binary search tree, but can it stop interrupting me every three seconds? Didn't think so.

ChatGPT out here asking for your .env file like it's NBD. You know, that sacred text file containing your API keys, database passwords, OAuth secrets, and basically everything that would make a security engineer have a panic attack. The confidence with "I'll fix it exactly 👍" is what really sells it though. Sure buddy, just gonna casually send over the keys to the kingdom so an LLM can debug my environment variables. What could possibly go wrong? Next thing you know, your AWS bill is $47,000 because someone's mining crypto with your credentials. The "BTW" in the header really captures that casual, almost apologetic tone of ChatGPT asking you to commit the cardinal sin of sharing secrets. Hard pass, my dude.

You: "Hey Claude, can you help me fix this one small bug?" Claude: "Sure! Let me just refactor your entire application architecture, delete half your files, rewrite your database schema, and while I'm at it, why don't we switch to a completely different tech stack?" AI coding assistants have this adorable tendency to go full scorched-earth when you ask for the simplest thing. You want a typo fixed? Cool, here's a complete system redesign. The horror on that face? That's you realizing Claude just wiped out three weeks of work because you forgot to be extremely specific about what "simple question" means. Pro tip: Always check what files Claude is about to touch before hitting accept. Your git history will thank you.

Normal humans click YouTube links with the carefree abandon of someone who's never heard of tracking algorithms. Meanwhile, programmers are over here performing digital forensics before every click, paranoid that the recommendation algorithm is secretly building a psychological profile. The incognito tab isn't just a browser feature—it's our tinfoil hat against the machine learning overlords. Because nothing says "professional paranoia" like treating a cat video recommendation like a potential security breach.