Andrej Karpathy on coding with AI assistants:

Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write… in words. It hurts the ego a bit but the power to operate over software in large “code actions” is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I’d expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent.

As a generalist, I relate to this. I’m increasingly doing things that I otherwise would’ve never dreamt of. In any cases, LLMs are like a jetpack for my brain.

Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro).

I’m not a coder, but I’ve become wholly reliant on LLMs for writing through voice, and I’ve kinda lost the desire to sit and type now. My first instinct is always to use the voice capabilities of LLMs. That said, I’m putting in some checks by carving out spaces for me to do the old-school manual and disgusting writing, voice typing, and LLM-assisted writing.

Having said that, the cognitive price of using LLMs is something I’m worried about the most, but it’s also inevitable that they will change the old ways of doing things in a lot of domains. It’s both exciting and terrifying to live through a technological paradigm shift in which the old ways of doing things are, in a lot of cases, not only inefficient but a liability. For over 30 years of my life, nothing really changed in terms of how I learn and do things. But in a span of 3 years, LLMs have changed so many things, even for a non-technical person like me, that I often feel disoriented.

Atrophy. I’ve already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it.