Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- Time will tell.by tomwuu
- > It helps to know that LLMs don’t “reason”. They predict ..
Semantics. Prediction is the training objective. The ability to reason can be, and very arguably is, an emergent property of that.
by theteapot - No. You're simply wrong.
- Why would "reasoning" be an emergent property of prediction?by slopinthebag
- Maybe it looks like reasoning, and maybe that's enough for some.by complex_pi
- By that standard, human brains don't either. Our externalizations of concepts like language or symbolic structure allow us to do so. In the parlance of our times, we built our own reasoning harnesses because our intuition lead us to do so.by TheWrongGuy
- Even if that was true, you'd have to still prove it has emerged.by jayd16
- I've never found these discussions to be all that useful, because it's hard to define what conditions are sufficient to say something is "thinking" or "reasoning". It just ends up being circular and metaphysical arguments.
That being said, current generation LLMs do have issue, it's more productive to talk about those and their impact on real tasks (long term memory, continual learning, tokenization, context rot, reversal curse, etc)
by nsingh2 - "They’re foundationally incapable of always and consistently preventing prompt injection attacks. “Alignment work”, safety harnesses, and sandboxes all help to add barriers against the worst, but there are fundamental gap" ...
They seem to be very good at a lot of rudimentary best practices, more so than humans, but more accurately - if you run and audit pass with specific instructions ... they're very good at that.
I mean - it's what they're the best at which is applying 'fuzzy heuristics' in a mechanical way. If can describe issues concisely, the patterns, the styles, the rules then LLMs can very mechanistically and methodologically grind through them.
I don't even see how this is controversial - without getting into 'what their reasoning means' - we can all agree that their synthetic reasoning is pretty good at narrow scales, and they've been 'trained by compilers' and are extremely good at spotting common patterns.
If you back that up with a lot of tokens ... they excel.
Designing architecture, that's difficult, but hammering away at all the 'known-knows across a system' especially to identify things ... they're pretty good at that.
by bluegatty - But do the same SWE fundamentals apply if the one doing the programming is many times smarter than us?by amelius
- Great. We are pretending that an LLM can think, and has intelligence. I thought the marketing wasn't supposed to work on us.by platevoltage
- LLMs are not smart in any way, but they are a weird kind of effective thats very different.
- I built a harness from scratch prior to trying any of the ones out there, so I knew how it work in a real way. I QUICKLY understood that the biggest issue with getting my shit done is that _my_ inputs are the untrusty ones. How many time do you hit backspace in a day?by RALaBarge
- An interesting question. We (humans) care about maintainability because the codebase will be adapted by teams of us for many years, based on new feature and bug fixes.
Rewriting from scratch is basically never an option after a certain amount of time.
Maybe the machines could just start from scratch each time and come at maintainability from a totally different angle
by baliex - I think SWE fundamentals matter - because it will be a long time before software is a closed system.
And the problem is that as long as humans are in the loop building software that dynamic will have to be maintained.
We use Loki for logging at work. There’s certain types of queries it just doesn’t support. And so the question becomes -
Do you change logging providers Adapt to Loki’s capabilities Create a third layer / tiered storage.
And each of those decisions have multiple downstream consequences. It’s not that LLMs can’t make those decisions per say, it’s that
What does an LLM do when five different people ask for a system optimized to do five different things.
It could figure it out itself, but like I don’t think that’s how the human software contract works.
by mikgp - It wouldn't make sense to confuse attributes of the software with attributes of the programmer.
Otherwise, one could simply declare one's IQ in a const somewhere, and have all unit tests follow the form: if the programmer's IQ is high enough, then the method under test is likely correct.
by mrkeen - > In the past year, agent harnesses crossed the “can it be done” rubicon.
Brother, I'm still in "Can you get it right?"-mode. What am I doing wrong? (Rhetorical, but advice welcomed).
by hirvi74 - There are two 'camps' with respect to AI.
One camp already knows that Neural Nets don't work and are a dead end.
The other camp hasn't yet figured out that Neural Nets don't work, but are convinced that they do (or eventually will), because they think everything always improves over time in a linear fashion.
by dosisking - What is “it” specifically and what languages are you using?by MattGaiser
- I'd be happy to screenshare with you if you like, we can work on something trivial or open source. Half an hour should be more than enough to see whether you're doing anything obviously self-sabotaging.by jaggederest
- You're appealing to ambiguity. All you've said is you have failed—how is anyone supposed to know what went wrong?by mw888
- I’ve found some success is small projects, with limited scope, in a greenfield.
I’m terrified to attempt agentic anything in the repo my job actually cares about. I triggered it once by accident, when the agent was first rolled out and enabled by default… it broke everything. Now I just use ask mode, and even that is wrong half the time, and once it goes wrong it just keeps getting worse.
I saw a post from Dave Plumber who vibe coded up a new cross platform task manager. He said his spec document for the AI was 107 pages long. So maybe what I’m doing wrong is not giving the AI a literal novel of spec.
by al_borland - Tell it to use red/green TDD and start things off with an already configured test suite, maybe with a single test that asserts 1+1==2.
Make sure it know how to run the tests before it starts writing any additional code.
Then set it a clear goal.
by simonw - LLM is the new Excel
- hey that's my line.by Krei-se
- The article says what many here like to hear, but in my opinion the core arguments are false.
> Making software debuggable, maintainable, layered, and composable – that’s still quite a trick
Not really. I have been working on a mobile app for months, and I stopped even glancing at the code about two months ago.
150k LOC, around half of that in tests, and the AI still has no problem maintaining the code on my behalf.
Debuggable? It can add extensive instrumentation in seconds.
None of this requires expertise, prompting, or mention of TDD. It's the default.
Frankly I do not believe the author tried developing a large codebase fully agentic and without reviewing the code. I believe many here look at the code produced, deem it substandard, and go hands on.
> They’re foundationally incapable of always and consistently preventing prompt injection attacks
From Anthropic's article about the Auto mode:
> We commissioned an evaluation from a third party, Trajectory Labs, who tested different models within the latest publicly available versions of Claude Code and Codex as of July 17th 2026.1 They tested 72 indirect prompt injection scenarios held out from Anthropic
> In this evaluation, none of the 720 attack attempts succeeded against Claude Fable 5, Opus 5, or Sonnet 5 running auto mode. On the other hand, 5.83% of the attacks succeeded against GPT-5.6 Sol running Codex's Auto-review mode. Notably, this is greater than the 0.09% average attack success rate against our latest models running in bypassPermissions mode without additional safeguards. The tests showed a 19.03% attack success rate against GPT-5.6 Sol when running in Full Access mode
I'm sure someone is going to reply with how they do not trust Antrophic's research, but lacking other data, prompt injection appears to be largely solved already.
by user43928 - > I'm sure someone is going to reply with how they do not trust Antrophic's research, but lacking other data, prompt injection appears to be largely solved already.
You do not solve "losing russian roulette" by adding more empty chambers. One bullet in 600 chambers is still one bullet too much. How about we don't play the stupid game in the first place?
by chmod775 - I tried this recently and the results were total banana cakes. They couldn't even make changes to Pong without breaking it.by andai
- > prompt injection appears to be largely solved
That's about as correct as saying cyber security is largely solved by referring to a "benchmark" that a particular virus scanner is able to detect and prevent infection with 720 known computer viruses. All this shows is that the model has been fitted to the benchmark, not that it is hardened against any conceivable way of prompt injection.
And regarding vibe-coding a mobile app with zero users, probably zero monitoring and zero everything like all the other vibe-coded apps that have zero users. Sure you can do that but what is the point?
by trixn - “Debuggable? It can add extensive instrumentation in seconds.
None of this requires expertise, prompting, or mention of TDD. It's the default.”
I’m pretty sure it takes some level of expertise just to use the term “instrumentation” correctly in a sentence.
by mikgp