AI assisted coding, when you can't get away with vibes. I've felt a mixed reception when talking to people integrating AI in their programming workflows. It has personally worked well for me, and I wrote about what I've tried so far.

"LLMs are just tools" never rang true to me. I'm reminded of a plot point in Becky Chambers' book where intelligent factory bots chose to walk away from serving humanity. I'd not equate LLM consciousness to human consciousness, but it makes sense to acknowledge and study it.

grateful for my low-quality education where teachers could not correctly explain the basics of anything right. it pushed me to cultivate good epistemics and build truth-seeking habits early on. it also forced me to never trust authoritative figures, many are very incompetent.

getting seen on hacker news is still a net positive in 2025. while everyone thrashes and nitpicks in the comments publicly, lots of interesting and nice people reach out privately!.

There's a more. Link to blog:

This is a good example of the weirdness of AI.

Why Does AI Feel So Different?. An enjoyable read from my colleague @SrihariSriraman. We've been talking about why this disruption feels different from other recent technological disruptions and Srihari captured a lot of that really well in this post.

ai won't solve a lot of the world's problems actually.meditation will. gautama et al >>> vaswani et al.

I don't disagree with Paras on needing fresh ideas btw, but it's weird to see some people on here suddenly bearish on scaling even though it doesn't seem to be objectively doing worse than it was many months ago. I don't see how the GPT-5 data point should change anything.

can those knowledgeable about LLM internals explain to me why I'm suddenly seeing "LLMs have plateaued" type posts *now*?. what changed? GPT-5 is on trend with RL scaling?. what should GPT-5 have done to convince people that LLMs haven't plateaued?. it's only been ~3mo since o3!.

the whole subreddit is actually very dystopian.

that whole reddit thread really made me update on the severity of chatgpt psychosis. very grim to see this. expect to see new heights of brainrot being achieved over the next few years.

RT @agrawalshagun10: Read @AtharvaRaykar post on why you should read I am a Strange Loop, which finally pushed me to read it. https://t.co/….

henceforth proposing that everyone not named douglas hofstadter stops making and using analogies.

analogies are like buses—they are clunky and rarely work well.

RT @AdvaitRaykar: Want to work on cutting edge applied AI?.Have a product engineering mindset? .Want to work on an impactful product, that….

My brother is looking for a high-ownership Product Engineer. You will build AI agents that help responsible sourcing teams manage the busy work in their supply chains. The team is fantastic, and you'll learn a lot here. It's fully remote. Do apply!.