We tried hard to make Devin work for us, But it didn't | Hacker News Reader

We tried hard to make Devin work for us, But it didn't

twitter.com

22 pointsamrrs6 comments

As someone who is wading through a bloated Devin PR right now, this article resonates with me. A direct link might be useful to some: [1]

1: https://www.answer.ai/posts/2025-01-08-devin.html

I have been surprised by the strength of Devin at the beginning of my team's use of it, but lately it has been a letdown. I don't know that we are ready to cancel yet but Devin faces some growing headwind for me, at least.

There was an earlier hn post on this: I wonder how effective ai coders are at translating libraries from one language to another. It’s plausible that they could be good at this I think it’s useful: anything from porting c++ code to rust to porting ecosystem libraries to languages that aren’t java/python/typescript

TL;DR: People with more money than common sense are surprised that LLMs are bad at coding despite not having there been a breakthrough that allows these models to think logically, which is a prerequisite for programming. They take one entire month to come to the conclusion that transformers are only good at producing hello world-level code.

Not only did they waste their money, but also a ton of time (and thus more money). Only to reiterate what is already painfully obvious to anyone who has worked with these tools for about five minutes (granted, you need to be an experienced programmer in order to see this -- but it's painfully obvious nonetheless).

Daily reminder that if Devon actually worked, it would be doing your job right now. The only reason you or your irrelevant company get to subscribe to it is because it's shiny crap and we're in the middle of a gold rush. Or you wouldn't even be able to afford it (what with being homeless and all).

A bit presumptuous to go into a rant that boils down to calling Jeremy an unskilled idiot. Jeremy has continuously used his programming experience and common sense to make more of an impact via his open source libraries, his courses on deep learning for coders, and language modeling research in general, than like.. 99.9% of programmers. It's basically comical how much you have misjudged him.

Then why was he surprised to find out that Devon is basically just ChatGPT? He should know first-hand what is is.

You're chatgpt.

Reply on news.ycombinator.com