The funny thing about AI discourse right now is how often someone’s prediction about what AI is going to mean for software development accidentally reveals what they think matters when building quality software. 😇😂
“We don’t need unit tests anymore.”
Translation: I always thought writing tests was just a pointless chore.
“In the future, it won’t matter whether a feature uses 100 lines or 5,000 lines of code.”
Translation: Isn’t simplicity a bit overrated?
“Clean code won’t matter once agents are writing everything.”
Translation: I thought “it works” was the end of the code-quality discussion.
“Documentation won’t matter because the AI can just inspect the code.”
Translation: I have always considered archaeology an acceptable onboarding strategy.
“Technical debt won’t matter because agents can fix it later.”
Translation: I have discovered a revolutionary new strategy called “make it someone else’s problem.”
“Experienced engineers won’t matter much because everyone will have the same AI tools.”
Translation: Access to tools is definitely the same thing as having professional judgment.
“Product knowledge won’t matter because the agent can read the tickets.”
Translation: Apparently, everything important has always been documented in easily discoverable Jira tickets.
“We don’t need to worry about upgrades. AI can handle them later.”
Translation: Future me is always on call.
“All these skills involved in building maintainable, reusable, robust software are becoming obsolete.”
Translation: I never understood why we were doing any of this in the first place. 😎