Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Less human AI agents, please

nial.se

Post reply on HN
Print view View original

51–60 of 184 posts

Re: Less human AI agents, please

#51

Post by felipeerias »

gregates wrote: post #15

The version of this I encounter literally every day is: I ask my coding agent to do some tedious, extremely well-specified refactor, such as (to give a concrete real life example) changing a commonly used fn to take a locale parameter, because it will soon need to be locale-aware. I am very clear — we are not actually changing any behavior, just the fn signature. In fact, at all call sites, I want it to specify a def…

Claude 4.7 broke something while we were working on several failing tests and justified itself like this:

> That's a behavior narrowing I introduced for simplicity. It isn't covered by the failing tests, so you wouldn't have noticed — but strictly speaking, [functionality] was working before and now isn't.

I know that a LLM can not understand its own internal state nor explain its own decisions accurately. And yet, I am still unsettled by that "you wouldn't have noticed".

Re: Less human AI agents, please

#52

Post by zingar »

nialse wrote: post #43
zingar wrote: post #35

Earlier quoted context omitted.

The article makes it seem like the author expected this without emptying context in between, which does not yet exist (actually I'm behind on playing with Opus 4.7, the Anthropic claim seems to be that longer sessions are ok now - would be interested to hear results from anyone who has).

That is probably the next step, and in practice it is much of what sub-agents already provide: a kind of tabula rasa. Context is not always an advantage. Sometimes it becomes the problem. In long editing sessions with multiple iterations, the context can accumulate stale information, and that actively hurts model performance. Compaction is one way to deal with that. It strips out material that should be re-read from…

I have the same experience of reversing intentional steps I've made, but with Claude Code. I find that committing a change that I want to version control seems to stop that behaviour.

Long context as disadvantage is pretty well discussed, and agent-native compaction has been inferior to having it intentionally build the documentation that I want it to use. So far this has been my LLM-coding superpower. There are also a few products whose entire purpose is to provide structure that overcomes compaction shortcomings.

When Geoff Huntley said that Claude Code's "Ralph loop" didn't meet his standards ("this aint it") the major bone of contention as far as I can see was that it ran subagents in a loop inside Claude Code with native compaction; as opposed to completely empty context.

I do see hints that improving compaction is a major area of work for agent-makers. I'm not certain where my advantage goes at that point.

Re: Less human AI agents, please

#53

Post by nialse »

jansan wrote: post #20

I disagree. I wan't agents to feel at least a bit human-like. They should not be emotional, but I want to talk to it like I talk to a human. Claude 4.7 is already too socially awkward for me. It feels like the guy who does not listen to the end of the assignment, run to his desks, does the work (with great competence) only to find out that he missed half of the assignment or that this was only a discussion possible s…

Interesting. When I code, I want a boring tool that just does the work. A hammer. I think we agree on that the tool should complete the assignment reliably, without skipping parts or turning an entirely implementable task into a discussion though.

Re: Less human AI agents, please

#54

Post by anuramat »

nialse wrote: post #14
js8 wrote: post #12

A very human thing to do is - not to tell us which model has failed like this! They are not all alike, some are, what I observe, order of magnitude better at this kind of stuff than others. I believe how "neurotypical" (for the lack of a better word) you want model to be is a design choice. (But I also believe model traits such as sycophancy, some hallucinations or moral transgressions can be a side effect of trainin…

Codex in this case. I didn't even think about mentioning it. I'll update the post if it's actually relevant. Which I guess it is. EDIT: It's specifically GPT-5.4 High in the Codex harness.

weird, for me it was too un-human at first, taking everything literally even if it doesn't make sense; I started being more precise with prompting, to the point where it felt like "metaprogramming in english"

claude on the other hand was exactly as described in the article

Re: Less human AI agents, please

#55

Post by gregates »

solumunus wrote: post #42
gregates wrote: post #27

Earlier quoted context omitted.

Indeed! You would think it would have some kind of sense that a commit that obviously won't compile is bad! You would think. It would be one thing if it was like, ok, we'll temporarily commit the signature change, do some related thing, then come back and fix all the call sites, and squash before merging. But that is not the proposal. The plan it proposes is literally to make what it has identified as the minimal cha…

On my compiled language projects I have a stop hook that compiles after every iteration. The agent literally cannot stop working until compilation succeeds.

In the case I described no code changes have been made yet. It's still just planning what to do.

It's true that I could accept the plan and hope that it will realize that it can't commit a change that doesn't compile on its own, later. I might even have some reason to think that's true, such as your stop hook, or a "memory" it wrote down before after I told it to never ever commit a change that doesn't compile, in all caps. But that doesn't change the badness of the plan.

Which is especially notable because I already told it the correct plan! It just tried to change the plan out of "laziness", I guess? Or maybe if you're enough of an LLM booster you can just say I didn't use exactly the right natural language specification of my original plan.

Re: Less human AI agents, please

#56

Post by cadamsdotcom »

gregates wrote: post #15

The version of this I encounter literally every day is: I ask my coding agent to do some tedious, extremely well-specified refactor, such as (to give a concrete real life example) changing a commonly used fn to take a locale parameter, because it will soon need to be locale-aware. I am very clear — we are not actually changing any behavior, just the fn signature. In fact, at all call sites, I want it to specify a def…

Make it write a script with dry run and a file name list.

You’ll be amazed how good the script is.

My agent did 20 class renames and 12 tables. Over 250 files and from prompt to auditing the script to dry run to apply, a total wall clock time of 7 minutes.

Took a day to review but it was all perfect!

Re: Less human AI agents, please

#57

Post by anuramat »

lexicality wrote: post #7

The entire point of LLMs is that they produce statistically average results, so of course you're going to have problems getting them to produce non-average code.

they (are supposed to) produce average on average, and the output distribution is (supposed to be) conditioned on the context

Re: Less human AI agents, please

#58

Post by nialse »

cadamsdotcom wrote: post #56
gregates wrote: post #15

The version of this I encounter literally every day is: I ask my coding agent to do some tedious, extremely well-specified refactor, such as (to give a concrete real life example) changing a commonly used fn to take a locale parameter, because it will soon need to be locale-aware. I am very clear — we are not actually changing any behavior, just the fn signature. In fact, at all call sites, I want it to specify a def…

Make it write a script with dry run and a file name list. You’ll be amazed how good the script is. My agent did 20 class renames and 12 tables. Over 250 files and from prompt to auditing the script to dry run to apply, a total wall clock time of 7 minutes. Took a day to review but it was all perfect!

Asking for code to manipulate the AST is another route. In python it can do absolute magic.

Re: Less human AI agents, please

#59

Post by comrade1234 »

gregates wrote: post #15

The version of this I encounter literally every day is: I ask my coding agent to do some tedious, extremely well-specified refactor, such as (to give a concrete real life example) changing a commonly used fn to take a locale parameter, because it will soon need to be locale-aware. I am very clear — we are not actually changing any behavior, just the fn signature. In fact, at all call sites, I want it to specify a def…

You can do that in IntelliJ in about 15 seconds and no tokens...

Re: Less human AI agents, please

#60

Post by nialse »

chrisjj wrote: post #39
prymitive wrote: post #25

Earlier quoted context omitted.

That’s my daily experience too. There are a few more behaviours that really annoys me, like: - it breaks my code, tests start to fail and it instantly says “these are all pre existing failures” and moves on like nothing happened - or it wants to run some a command, I click the “nope” button and it just outputs “the user didn’t approve my command, I need to try again” and I need to click “nope” 10 more times or yell a…

> it breaks my code, tests start to fail and it instantly says “these are all pre existing failures” and moves on like nothing happened Reminds us of the most important button the "AI" has, over the similarly bad human employee. 'X' Until, of course, we pass resposibility for that button to an "AI".

The other day Codex on Mac gained the ability to control the UI. Will it close itself if instructed though? Maybe test that and make a benchmark. Closebench.
Post reply on HN