How AI Is Collapsing the Distance Between Ideas and Software
Hello Reader, Michael here from Notion Panda. It's been a while since I shared on this newsletter but I am excited to be back. Although I have been writing more frequently with The Flying Sage Journal, I am desiring a place to share more about business, tech and entrepreneurship and so therefore, my intention with thisnewsletter is to provide value to entrepreneurs and practitioners in the community building and healing spaces. Today I want to talk about vibe-coding and how software is being democratized.
Going into university, I thought I wanted to be a programmer. I studied some computer science, took the core courses, wrestled with problem sets and passed—though not without difficulty. The emerging psychedelic industry eventually pulled my attention elsewhere however, and because I didn’t thrive in my CPSC classes, I determined that software design was not for me.
But now I am being forced to reconsider this conclusion. As AI-powered development and vibe coding tools accelerate at what seems to be an absurd pace, I find myself pondering a question that feels increasingly relevant:
If someone wanted to learn how to program now, where should they actually focus their time?
From Bits to Meaning: A Brief History of Abstraction
One of the most formative courses I took at UBC was CPSC 121, a class focused almost entirely on first principles: how computers work at the most basic level.
We learned about:
binary (1s and 0s)
logic gates
memory
boolean algebra
finite state machines
sequential instruction execution
how electrical signals become computation
From there, I took a few more classes and moved up the abstraction ladder:
Machine code – raw instructions executed by the CPU
Assembly – human-readable representations of machine instructions
Compiled languages like C and C++, which translate down to machine code via compilers
Higher-level languages like Python, JavaScript, and Ruby, which abstract memory management, pointers, and hardware details almost entirely
Each of these layers exists to reduce cognitive load.
Very few modern programmers need to think about registers or instruction pipelines. Even fewer need to manually manage memory the way C programmers once did. Instead, they operate in environments where entire systems are built from libraries, frameworks and APIs that hide enormous complexity beneath the surface.
And that abstraction has only accelerated.
The New Layer: AI as a Programming Interface
Tools like Replit, Lovable, v0, Cursor, and GitHub Copilot represent something fundamentally new.
They don’t just abstract syntax.
They abstract intent.
You can now sit down, describe what you want in natural language, and collaboratively build working software with an AI companion. Databases, authentication, UI components, backend logic—all scaffolded in minutes.
I recently built a piece of internal software for the Deep Work Detoxprogram that tracks participant goals across cohorts. I took what I was managing in a spreadsheet, and in a few hours, created a reliable working prototype.
Five years ago, that same tool would have required:
backend architecture
database schema design
frontend logic
authentication flows
deployment infrastructure
The Deep Work Detox app I vibe-coded
A Useful Analogy: Arithmetic and Calculators
This shift reminds me of a familiar argument in education:
“You still need to learn arithmetic, even though you’ll always have a calculator.”
Most people don’t manually calculate long division in daily life. But learning arithmetic develops intuition, number sense, and problem-solving ability.
Programming feels similar.
Understanding binary, memory, algorithms, and data structures undoubtedly makes someone a stronger programmer. Someone who understands Python will generally prompt and debug AI tools more effectively than someone with zero exposure.
But here’s the uncomfortable question:
How large is that advantage—and how long will it last?
If AI tools continue to improve, the gap between “experienced programmer” and “thoughtful beginner” may shrink dramatically.
What Still Matters (And Probably Always Will)
Even as abstraction increases, certain skills remain non-negotiable:
1. Systems Thinking
Knowing how components interact and how data flows through a system.
2. Critical Reasoning
Being able to evaluate outputs, spot inconsistencies and ask better questions.
3. Iteration and Debugging
AI produces drafts, not perfection (yet). Someone still has to notice when things break.
4. Conceptual Literacy
You don’t need to master assembly—but understanding what a database is, what an API does, and why latency exists still matters.
In other words, thinking like a programmer matters more than writing code like one.
Will Programming Languages Become Historical Artifacts?
This leads to another question:
Will programming languages eventually become something people learn mostly for:
historical understanding
intellectual enrichment
niche optimization
or sheer enjoyment
Much like Latin or advanced mathematics?
It’s not that they’ll disappear—but they may no longer be the primary gatekeepers to software creation.
The barrier to entry is already approaching zero.
If you can:
articulate goals clearly
reason about structure
prompt effectively
and iterate thoughtfully
You can now build software.
So Where Should Someone Focus Today?
If you’re starting now, my honest answer is this:
Learn how software systems work conceptually, not just syntax
Learn one high-level language (Python or JavaScript are still great) to build intuition but have AI support you in learning it
Spend time collaborating with AI tools, not competing with them
Practice designing systems, not memorizing frameworks
Understanding lower layers still provides leverage—but it’s no longer mandatory for participation.
The Vibe Coding Era
What’s happening isn’t the death of programming.
It’s the migration of value up the abstraction stack.
Just as most people don’t need to understand transistors to use a smartphone, fewer people will need to understand code to build software.
The skill of the future programmer isn’t keystrokes.
It’s judgment.
As long as you can think clearly, ask good questions, and remain curious, you can participate in building the digital world.
Notion Panda helps entrepreneurs, holistic practitioners, community builders, and small business owners streamline operations through intuitive templates and hands-on consulting.