DEV Community

Sachin Kizhakkepurath
Sachin Kizhakkepurath

Posted on

After 3 months of AI doomscrolling, I realized most developers are worried about the wrong things.

For the last few months, I've been reading AI papers, watching demos, testing ChatGPT, Claude, Cursor, Copilot, and generally trying to figure out what AI actually means for developers.

Like a lot of people, I started from a place of anxiety.

Every week there seemed to be a new headline:

  • AI writes code
  • AI builds apps
  • AI replaces engineers
  • One developer can now do the work of ten

The more I dug into it, the more I noticed something interesting.

AI is incredibly good at generating answers.

It's much less good at figuring out what question should be asked in the first place.

That's the part many discussions miss.

In my day-to-day work, the hardest problems are rarely about writing code.

They're usually things like:

  • Understanding vague requirements
  • Figuring out what stakeholders actually want
  • Deciding between multiple imperfect solutions
  • Navigating technical debt
  • Balancing business constraints with engineering quality

These are fundamentally context problems.

AI can help with them, but it doesn't own them.

Another thing I realized is that experienced developers often underestimate how much domain knowledge they've accumulated.

When I look at a piece of code, I'm not just reading syntax.

I'm thinking about maintainability, edge cases, customer impact, previous incidents, team conventions, performance tradeoffs, and future changes.

A language model can generate code.

It doesn't have years of accumulated context about why a particular decision matters.

Ironically, AI has made me value human skills more, not less.

Communication.

Judgment.

Problem framing.

Understanding people.

Those things have become more important because the mechanical part of coding is getting cheaper.

My current view is that developers who learn to use AI effectively will probably become much more productive.

But I don't think the biggest risk is AI replacing developers.

I think the bigger risk is developers assuming that writing code was the entirety of their value.

Curious how other people are thinking about this.

Have AI tools changed your view of what makes a good developer?

Further Reading

I eventually turned my research into a short ebook called From Fear to Fluent.

If you'd like a deeper dive into:

  • How LLMs actually work
  • What AI can and cannot replace
  • Using AI as a pair programmer
  • Future-proofing your career as a developer

You can check it out here:

👉 Get the ebook

Top comments (0)