Hacker News Reader

Comment by efsavage | Hacker News Reader

It was "writing 90% of the code", which seems to be pretty accurate, if not conservative, for those keeping up with the latest tools.

View on HN

> which seems to be pretty accurate

It's not, even by his own citing: https://www.youtube.com/watch?v=iWs71LtxpTE

He said that this applies to "many teams" rather than "uniformly across the whole company".

Yes, those using the tools use the tools, but I don't really see those developers absolutely outpacing the rest of developers who do it the old fashioned way still.

I think you're definitely right, for the moment. I've been forcing myself to use/learn the tools almost exclusively for the past 3-4 months and I was definitely not seeing any big wins early on, but improvement (of my skills and the tools) has been steady and positive, and right now I'd say I'm ahead of where I was the old-fashioned way, but on an uneven basis. Some things I'm probably still behind on, others I'm way ahead. My workflow is also evolving and my output is of higher quality (especially tests/docs). A year from now I'll be shocked if doing nearly anything without some kind of augmented tooling doesn't feel tremendously slow and/or low-quality.

it’s wild that engineers need months or years to properly learn programming languages but dismiss AI tooling after one bad interaction

Motivated reasoning combined with incomplete truths is the perfect recipe for this.

I kind of get it, especially if you are stuck on some shitty enterprise AI offering from 2024.

But overall it’s rather silly and immature.

I think inertia and determinism play roles here. If you invest months in learning an established programming language, it's not likely to change much during that time, nor in the months (and years) that follow. Your hard-earned knowledge is durable and easy to keep up to date.

In the AI coding and tooling space everything seems to be constantly changing: which models, what workflows, what tools are in favor are all in flux. My hesitancy to dive in and regularly include AI tooling in my own programming workflow is largely about that. I'd rather wait until the dust has settled some.

totally fair. I do think a lot of the learnings remain relevant (stuff I learned back in April is still roughly what I do now), and I am increasingly seeing people share the same learnings; tips & tricks that work and whatnot (i.e. I think we’re getting to the dust settling about now? maybe a few more months? definitely uneven distribution)

also FWIW I think healthy skepticism is great; but developers outright denying this technology will be useful going forward are in for a rude awakening IMO

That's not even close. The keyboard is writing 100% of my code. They keyboard is not replacing me anytime soon.

If you added up all the code written globally on Dec 3 2025, how much do you think was written by AI and how much was clacked out on a keyboard?

Reply on news.ycombinator.com