Will AI and automation set us free, or replace us?

Back when I worked as an RPA consultant, I heard the same promise everywhere: “Automate repetitive and mundane work, and people will move on to more valuable tasks.”

But what I actually observed in the field was different. In many cases, automation didn’t “elevate” work—it simply removed it. If a bot can run 24/7 with minimal intervention, organizations often treat that as a replacement, not a redesign of human work.

Today, I work as a technical coach in a multi-agent building course (LangChain, RAG, and more). Part of my job is removing friction for learners—debugging and fixing complex python code so they can focus on learning.

In that context, Claude Code has been a real “best friend.” I diagnose the problem, give clear instructions, and it executes the changes quickly. What impressed me most is its ability to update many files across an entire project folder just by pointing to the path—massive time savings.

For Office work, I used to bounce between Word and web-based LLMs—copying and pasting content back and forth between Word and tools like Claude, Gemini, and ChatGPT.

But with Copilot (Agent mode under a commercial license), something changed: instead of editing files myself, I can instruct Copilot to edit the file for me—especially helpful when I’m not fully fluent in Word’s features and formatting.

Strength: You can describe the outcome in natural language (restructure a section, turn notes into a table, tighten wording), and Copilot acts like an agent—actually making the changes for you inside Word.

Limitation (for now): It doesn’t yet feel as strong as Claude Code when it comes to making large-scale changes across many files in a folder. I expect that gap to close.

So let’s return to the original question: does automation free humans—or replace humans?

My answer: for rule-based repetitive tasks, replacement is the honest outcome—robots are simply better at that. But for ambiguous work (reports, planning, product decisions, human-centered applications), success depends on context alignment. That’s where human judgment still matters.

Pilot (human): Sets direction, prioritizes, and aligns stakeholders and context.
Copilot (AI): Executes the detailed work fast—drafting, formatting, summarizing, iterating—once the direction is clear.

To make this shift real, we need both: (1) a clear understanding of what today’s AI tools can and cannot do, and (2) the insight to identify the true essence of our work. In other words, we need both a Copilot and a Pilot.

AI and Copilot do not generate money. Money comes from human. It is your responsibility as pilot/director, to provide service to human to generate profit using Copilot, AI, or any other or none of these tools.