in a recent post on his social media account flagged a warning for coders, saying the artificial intelligence could render them as liabilities. "AI code capabilities are getting better and better. One criticism is "AI is now quite good at generating new code but it is not as good in helping us with a large existing code base" and that is valid. Now I have started to wonder if the answer is "If AI makes us 100x productive, why not rewrite the whole thing with AI help?" Of course it is not there _yet_ but perhaps not for long. Viewed that way, large existing code based may no longer be assets, they may be liabilities. Scary thought," said Vembu in a post on X.
Check his post here:
How did people react to Vembu's post:
"As a developer after using the Gemini 2.5 model, I feel like my job is at stake. What else I can do as a developer?" said one user. In response, Vembu replied "I felt the same way. Join the club!"
"I have seen so many good technical teams recommend this option at one stage or another. The risk was higher, and might take months to see the results. Now its faster, so i guess this risk is low now to do rewrites," said one user.
"I actually did something similar recently -- instead of trying to fix a spaghetti codebase I used AI to rewrite the entire repo from scratch. Gemini 2.5 Pro is really good at that," added another user.
"With the deepest respect, Mr. Sridhar — your insight pierces through convention. What was once seen as a strength — legacy code — may soon anchor us to the past. If AI can amplify human potential a hundredfold, why not reimagine from first principles? You’ve not just posed a question, but gently shattered an industry’s quiet assumption. This is not just a shift in technology — it’s a philosophical reset. Grateful for your relentless clarity in a world clouded by comfort," said another user.
