Prof Archie D'Souza
Since I started teaching logistics in 1989, I’ve opened
by telling my students ~ transportation has never been faster and easier than
it is today. What was true close to about four decades back is true today as
well. But AI is changing the way we think and act. The transport analogy can
be used for building code today. We have never been able to build faster. Open
an editor today, prompt an AI agent, and a fully functional app or feature
materializes in minutes.
The friction is gone. But with it went our greatest
safeguard: the thinking tax.
Has it?
Previously, building took time, money, and cognitive sweat.
That friction forced us to ruthlessly validate ideas before writing a single
line of code. Today, it’s dangerously easy to fall in love with how fast
we built something, blinding us to whether the world actually needed it in the
first place. No, I’m not attacking researchers, but today’s coder.
We’ve optimized for output, not outcome. Shipping ten
features a week means nothing if none of them solve a real human problem.
In a world where anyone can spin up a codebase in an
afternoon, your velocity is no longer your competitive advantage. Discernment
is.
AI can give you code, speed, and answers, but it cannot give
you conviction. It doesn't care why it's building. That
responsibility—and the choice of what problems are actually worth solving—rests
entirely on us.
The Illusion of Progress
The Reality: Open your editor, prompt an AI agent,
and watch a fully functional app or feature materialize in minutes. We have
never been able to build faster.
The Paradox: Yet, look around. How many AI-generated
tools are just solving trivial inconveniences while massive, systemic problems
remain untouched?
The Core Thesis: AI has democratized building,
but it hasn't democratized judgment. Speed without direction is just
accelerated wandering.
The Trap of Frictionless Code
The Old Guardrail: Previously, friction was our
friend. Building took time, money, and cognitive sweat. That forced us to
validate ideas ruthlessly before writing a single line of code.
The New Hazard: Frictionless creation removes the
"thinking tax." It’s dangerously easy to fall in love with how
fast we built something, blinding us to whether the world actually needed
it in the first place.
The Output vs. Outcome Fallacy: Shipping 10 features
a week means nothing if none of them move the needle for real human needs.
Redefining Value: From Coding to Curation
The Shift in Developer Role: We are transitioning
from "builders" to "architects of intent."
The Hardest Questions:
- Is
this problem actually worth solving?
- What
are the unintended second-order consequences of this solution?
- Are
we just automating noise?
The third question led me to write this.
The Real Moat: In a world where anyone can clone your
codebase in an afternoon, your competitive advantage isn't your velocity. It's
your taste, your empathy, and your ability to spot the right problem.
How to Build "Right" in the Age of AI
Slow Down to Speed Up: Use the time AI saves you on
execution to double down on user research, problem discovery, and critical
thinking.
Measure Impact, Not Output: Stop tracking lines of
code, PRs closed, or shipping velocity as badges of honour. Track whether you
are reducing friction for human lives.
Embrace Intentional Constraints: Artificially
introduce limits. Pretend you only have three prompts or a strict scope to
force yourself to focus on what truly matters.
Conclusion: The Human Advantage
- AI
can give you answers, code, and speed, but it cannot give you conviction.
It doesn't care why it's building. That responsibility rests
entirely on us.
- Closing
Thought: Let’s stop using AI just to build things faster, and start
using it to solve things that actually matter.
Stop using AI just to build things faster. Use the time it
saves you to figure out what's actually worth building.
