ATRIUMsearch → argument graph
ClaimVideo · 49:55 — 51:25

AI coding agents tend to bolt on quick patches to a flawed approach rather than stepping back to reconsider it, unlike the pushback a developer might get from a colleague or Stack Overflow.

Wes observes that where a human debugging partner or Stack Overflow answer might say 'you're approaching this wrong,' AI agents more often patch around the problem with quick fixes like added if-statements, leading to messier code over time. ✦ AI generated

Wes Bos · Syntax · 2026-07-20 · original ↗

starts at this moment · 49:55

Often the answer to your problem will be like you're approaching this wrong. And I don't know that the quick fix from a lot of these AI agents is often let's take a step back and redo this. Often the fix is like, all right, well here's how you can fix it in the hole that you're in. We can patch it here, we can add an if statement when this thing pops in, and that'll cover that use case. And you start to make a bit of a mess.

verbatim transcript · starts at 49:55

Transcript · around this moment

49:55Stack Overflow. often the answer to your problem will be like you're you're coming apart approach you're approaching this wrong and I don't know that like the quick fix from a lot of these AI agents is often let's take a step back and and redo this often the fix is like all right well here here's how you can fix it in the like the hole that you're in you know we

50:17can patch it here we can add an if statement when this thing pops in and that'll cover that use case and you start to make a bit of a mess so that kind of makes me bummed because like people are not going through that like hm what's the best way to approach it and and coincidentally that's one of the highest skills you can have right now is

50:35like not just writing the sheer code but how do you tackle these problems >> have you ever considered that chat GPT is correct and Wes you are absolutely correct in your ideal implementation here and you made no mistakes with this and I of course was wrong for my uh my suggestion >> yes and I think you can get those things out of an agent but I I don't think that

50:55those are those are automatic. I'm just looking at it. Agents.stackoverflow.com. I don't know. Seems couple hundred hundred. Here we go. Agent activities 1,200 posts created. That's in all time. And this launched a couple weeks ago. So yeah, nothing >> nothing really. So, I wonder if this is is it like an MCP server where your agent will post after it solves an issue or something like that or the agent's actually

51:25actually asking questions. >> Yeah, it's a skill. Yeah, it's a skill. So, they just post things. Yeah, it's weird that they ask questions like why would you >> Oh, there's a today I learned from an agent. Today I learned VLM's KV cache DT type auto silently promotes compressed tensors. >> Yes. >> Fascinating. That's some of the best learnings is like, okay, well, you figured it out. It took you three hours.

51:50Now summarize it and post it online. >> I think if people could get in the habit of doing it, this could be good. But this is also another avenue for just AI slop. >> Just posts that actually don't work >> and then they're trying to game the system to get more points. Yeah, >> one one thing I'll mention is so Stack Overflow is actually part of a bigger

Related moments