I recently sat down with Michael, a Senior Software Engineer at Square (and ex-Facebook) with over a decade of experience. I asked him the question that is on every parent and educator’s mind:
Will AI automate software engineering?
His answer was blunt: “I basically don’t write a single line of code without an AI companion... What we’ve done is basically replace coding labor. That is essentially a solved problem.”
This conversation wasn’t just about code; it was about the future of work, the trap of “purism,” and the one skill our children need to survive the shift.
The Death of “Coding Labor”
Michael made a distinction that is vital for educators to understand: the difference between Software Engineering and Coding Labor.
Coding Labor is the act of typing syntax—knowing where the semicolon goes or how to write a specific function from memory. That is gone. AI can “one-shot” entire applications now. It can generate a Windows interface replica in seconds—a task that used to take months.
Software Engineering, however, is about architecture, judgment, and “taste.”
Because AI has solved the labor portion, entry-level jobs are vanishing. The industry no longer needs juniors to “hammer the nails.” They need architects who know where to put the nails.
The Trap of the “Purist”
We discussed a fascinating trend: Engineers who are in denial.
There are “purists” who refuse to use AI on principle. They view it as “defiling the craft.” They want to write every line by hand to maintain total control.
Michael compares this to the shift from Assembly language to modern programming. Decades ago, you had to manually manage computer memory with zeros and ones. Today, modern languages handle that for you. AI is just the next layer of abstraction.
The harsh reality?
“You may have been a great engineer two years ago, but if you don’t leverage this tool now, you become less valuable than a worse engineer who knows how to leverage AI well.”
If you deny the productivity gains, you get left behind.
The New Superpower: High Agency
If AI handles the “how,” humans must master the “what” and the “why.”
Michael and I concluded that the most valuable trait in this new era isn’t IQ or memorization. It is High Agency.
High Agency is the ability to:
Identify a problem (Domain Expertise).
Determine if it can be automated (AI Literacy).
Execute the solution without waiting for permission or instructions.
It is about looking at a codebase (or a business, or a classroom) and seeing where efficiency can be created. It is about having the curiosity to say, “I don’t know this finance term,” and then talking to an AI until you do understand it, rather than just stopping.
How We Teach This
This brings us back to Someta.
Our goal isn’t just to help students get the right answer on a math test. If that was the goal, they could just use Photomath.
Our goal is to build Agency and Metacognition. We want students to have the “taste” to know when an answer looks wrong. We want them to have the curiosity to “debug” their own thinking processes.
As Michael said about raising his own daughter in this AI age:
“I want to cultivate curiosity, openness, and optimism. If you’re pessimistic, you’re closed off to learning.”
The future belongs to the curious.
Key Takeaways
Coding Labor vs. Engineering: Typing syntax is a solved problem; architectural judgment is where the value lies.
Don’t be a Purist: Refusing to use AI for “moral” or “craft” reasons is a career-limiting move.
Leverage is Key: A junior engineer with AI leverage can out-produce a senior engineer without it.
Agency > IQ: The ability to self-direct, learn “unknown unknowns,” and execute is the primary metric for future success.
