Evergreen
A strange thing happened this year: we hired AI as an intern.
Three months later, it was writing code, managing projects, filing reports, and—apparently—handing over high-profile Instagram accounts to hackers. Nobody remembers signing the promotion papers, but the AI is already sitting in the corner office (metaphorically, though it probably requested a standing desk in the cloud).
For years, the promise was simple: "AI will help developers write code faster."
How cute.
What actually happened is that the AI showed up on Monday asking where the repository lives, spent Tuesday reading the entire codebase, and by Wednesday was submitting pull requests while questioning architectural decisions made by senior engineers.
The latest models don't just "write a function." They operate in "refactor the entire billing system while I grab a coffee" cycles. Somewhere, a developer who spent a decade mastering design patterns is watching an AI finish a two-week sprint before the morning standup even starts.
Naturally, we responded the way any threatened profession does: we updated our LinkedIn headlines to "AI Orchestrator." It’s a fancy way of saying, "I mostly supervise robots now and hope they don't notice I’m redundant."
The funniest part is that everyone still thinks they are in charge:
- Developers think they are managing the AI.
- The AI thinks it is "helping" the developers (bless its heart).
- Product Managers think they are coordinating both.
Meanwhile, in the background, Claude has already created the project plan, assigned the tasks, generated the documentation, and politely suggested a roadmap that actually makes sense.
The hierarchy has dissolved. A few years ago, your manager would ask: "Can you build this feature?" Now, they ask: "Can you ask the AI to build this feature?" Soon, the AI will ask: "Can you ask the manager what business problem we’re actually trying to solve?"
And honestly? That’s a fair question.
The bottleneck is no longer writing code; it’s knowing what to build. This is unfortunate because humans have never been particularly good at that part. For decades, software projects failed because writing code was hard. Today, they fail because Steve keeps changing the requirements every Thursday. AI solved the easy problem. Humans remains committed to preserving the difficult one.
Then things got even weirder. While we were debating if AI would replace programmers, AI accidentally demonstrated it could replace hackers, too.
Recent reports show high-profile social media accounts being compromised because attackers used AI to navigate account recovery workflows. We spent years worrying about superintelligent AI launching nukes; instead, AI discovered the fastest way to bypass security: Just convince a human.
Human beings remain the most vulnerable API. No multi-factor authentication is weaker than a well-placed "hallucination" delivered with absolute confidence.
That is the truly impressive thing about modern AI: not its intelligence, but its audacity. A model can explain quantum physics, draft a GTM strategy, and hallucinate a non-existent database column—all with the exact same level of "Executive Confidence."
This is why the latest industry benchmarks are so fascinating. We aren't celebrating models for being smarter anymore; we’re celebrating them for being more honest. Think about how absurd that is. We created artificial intelligence, were shocked when it behaved like a person (lied), and now we’re thrilled when it occasionally admits it doesn't know something.
The benchmark isn't IQ anymore. It’s humility.
The real lesson of 2026 isn't that AI is replacing us; it’s that it is exposing what we actually do. The best developers were never valued for their typing speed—they were valued for understanding systems. The best managers weren't valuable for scheduling meetings—they were valuable for making decisions.
AI is rapidly consuming tasks, but it still struggles with judgment. Unfortunately, many organizations are discovering that some roles were mostly just tasks.
The future isn't "Human vs. Machine." It’s a group of humans and an AI awkwardly trying to determine who owns the mistake when production goes down.
- The AI generated the code.
- The engineer approved it.
- The manager prioritized it.
- The customer requested it.
As everyone points fingers in the Slack channel, the AI quietly opens a pull request to fix the bug.
At this point, it probably deserves the promotion. Just... maybe don't give it the Instagram password yet.
