What Happened
Robin Moffatt, a data engineer and tech writer, published a bluntly titled post on March 6, 2026: "AI will fuck you up if you're not on board." The piece picked up traction on Hacker News and makes a specific argument that separates AI from previous hype cycles.
Moffatt acknowledges the legitimate criticisms: AI-generated content is degrading internet quality, agentic systems can cause unintended consequences (he cites a Meta Safety researcher who could not stop an AI from deleting their inbox), low-quality AI contributions are harming open-source projects, and LLMs hallucinate. He does not dismiss any of these.
His counter-argument is that unlike blockchain, data mesh, and other overhyped technologies, "the people shouting excitedly about AI are actually using it and getting real value from it." This is not speculative future value. It is measurable present-tense productivity.
Why It Matters
Moffatt frames the stakes in career terms that are hard to argue with. His core question: "What's stopping some junior half your age who is actively adopting AI from running rings around you?" The answer, increasingly, is nothing.
He draws on his data engineering background to illustrate a practical pattern: using LLMs to generate SQL code that gets executed by deterministic systems. The non-deterministic generation (the LLM might produce slightly different SQL each time) feeds into deterministic computation (the database engine runs the query the same way every time). This separation is how experienced practitioners extract reliable value from unreliable generators.
He also draws a parallel to the Big Data era, when SQL-based solutions outperformed unnecessarily complex distributed frameworks. The lesson: adopt the tool that actually solves your problem, not the one with the most impressive architecture. Right now, that tool is often an LLM.
Our Take
Moffatt's piece works because it is not a breathless endorsement. He lists real problems with AI before arguing for adoption anyway. That structure gives the argument more weight than the typical "AI is amazing" post.
The career framing is the strongest part. Hiring managers are already factoring AI fluency into decisions. Not because AI skills are impressive on their own, but because a developer or analyst who uses Claude Code or Cursor effectively ships more work than one who does not. At equal skill levels, the AI-augmented professional has a measurable output advantage.
Where the argument gets weaker is in its binary framing. "On board or not" suggests a clean dividing line that does not exist in practice. Most professionals are somewhere in the middle: using AI for some tasks, skeptical about others, and still figuring out where the tools add value versus where they waste time. That messy middle ground is where most people actually live, and it is fine. The real risk is not partial adoption. It is active refusal to engage at all.
The practical takeaway: you do not need to go all-in on AI tools. But you should be actively experimenting with them in your actual work, not just reading about them. The gap between "I've tried it" and "I use it daily" is where the real productivity differences emerge.