My views about AI and AI assisted tools | Lakshmanshankar
Back to Blog
3 min read

My Views About AI and AI-Assisted Coding

I am a fullstack developer with 1.5 years of experience working on a small, user-focused application, so my views might differ from those who primarily work on enterprise software. It’s May 2025, and things are changing faster than ever. AI coding assistants are getting smarter, and companies have started adopting AI more seriously into their workflows.

I’ve heard both sides—some say AI will replace developers, others say that’s far off. I don’t fully agree with either. What’s clear is that change is happening, and I need to be ready for how the way we work is shifting.

How it all started:

I started using GitHub Copilot back in 2023 when I got free access through the GitHub Student Developer Pack. I still remember the first tab-completion—it felt amazing. Copilot could suggest anything from a few lines of code to entire small functions. After using it for a while, I noticed myself waiting for Copilot to auto-complete lines instead of thinking through the logic myself. That was when I decided to limit my usage to just a few lines at a time, rather than entire functions.

Recently, I started using Cursor. It has more features and offers better AI coding assistance. It can generate UI components easily and autocomplete functions more accurately. But this time, my thoughts are different. I believe I can’t skip this shift. I’m preparing for the transition because I see AI doing some things better than me and some things it still struggles with. Trying out a tool is one thing—realizing it’s now part of your daily workflow is something else.

Things AI Is Doing Better:

  1. Quickly generating UI components, especially with Tailwind CSS. Tools like v0 and Cursor match your style well.
  2. Writing small to mid-sized functions that save time without hurting quality.
  3. Creating solid test cases with minimal input.

Things AI Still Struggles With:

  1. Understanding user requirements and features.
  2. Breaking down requirements into actionable tasks. I believe this is the hardest part.
  3. Working with functions that are deeply integrated into the application.

Final Thoughts:

It’s definitely easier to do certain things now. I think of AI like a car for coding—it helps you move faster only if you know how to steer it. It won’t magically fix bugs or add features. Coding is still a valuable skill, but the way we work is changing fast. We’re shifting from typing everything line by line to guiding and orchestrating the AI to do the coding for us. I belive that is the real shift.


Questions:

  • Do you think AI is going to replace developers?
  • What do you think the future of coding looks like? Is it more about prompting?
  • What advice would you give to the younger generation currently studying computer science? What fundamental shifts should they prepare for?

You can mail your thoughts to me or on Twitter.

Lakshmanshankar © 2025