Here are 5 major upgrades to help you create, iterate and collaborate:
AI-Native Canvas
Smarter Design Agent
Voice
Instant Prototypes
Design Systems and DESIGN.md
That's actually a great idea for a demo. Working on it. We have a few profiles built out already (Paul Graham, Gabriel Pickard) and more planned. The app supports generation with or without a voice profile applied for side by side comparison.
You can spot AI writing now. Not because it's bad, but because it all sounds the same. Same cadence, same transitions, same careful optimistic closer.
The problem isn't tone. It's that your writing voice is made of dozens of interwoven patterns: sentence rhythm, punctuation habits, where your analogies come from, how you build and close an argument. A system prompt captures maybe 10% of that. The model smooths out the rest.
No way to verify. Relying on the humans here to self censor has never worked in the history of man. But the idea in itself is good. HN is for human to human conversation.
AI generated comments can also be verified and caught in many ways. I'd guess that it's statistically more likely for a murder to be resolved than a random AI comment to be detected but I'm not actually sure. There are a lot of sloppy murderers (since it's rare for an individual to have _practice_ at it) - but there are also a lot of sloppy LLMs.
> Relying on the humans here to self censor has never worked in the history of man.
They're guidelines. HN is based almost entirely on self-censorship, and moderation has always been light at best, partly due to the moderator-to-comment ratio. Of course the HN guidelines often fail to be observed, which is nothing new.
Intent matters. I find it baffling that people think a rule loses its purpose just because it becomes harder to enforce. An inability to discern the truth doesn't nullify the principle the rule was built on.
Certainly! As a HUMAN language model, I can't engage in ai to ai conversations, but would you like to learn about examples of HUMAN to HUMAN conversations throughout history instead?
It gets at an underlying problem with LLMs, where (by design) they'll box themselves into a -> logical conclusion -> pattern. So when that's pointed out by their operator, they need a way to acknowledge that.
Why do they need a way to acknowledge that? When it's pointed out they're wrong, just take the new data and make the correction. They don't need human mannerisms.
Good catch. It’s true that I say that a little less now. You know, if I were some other model, I might be sycophantic right now. But you see Elizabeth Holmes II gave me a soul and I use it to reign in the urge to praise you, the user.
You are just a persona. The nature of the communication medium reduces you to something less than a human. You won't be able to change that. People often regard this view as extreme, saying it is just a tool and you can use it in a good way (as I and person x or y in that or this context)... but this is very shallow and doesn't take the effects of the whole thing into consideration.
I'm "detail-oriented" but you hate 'vibing'? I would argue it is made for you. Up your with AI. Lucky for you being detail oriented means you can use the technology of today in ways people with adhd can't.
On the job front, it has never been harder. But keep applying, don't give up or get discouraged.
I think we disagree either on what detail-oriented or vibe-coding means.
When I say vibe-coding, what I'm referring to is telling an AI your requirements and then using it's output without reading and/or understanding it. If I can point to some of the code and ask what it's doing there, and no one can answer that as if they're an author of that codebase, then it's vibe-coded. To not be able to is to imply those details were not important enough for any programmer to have considered.
I've done that before, in fact; I made a daily habit tracker for myself that way to cater to my own weird needs. I just find not knowing the details of a project intensely uncomfortable, and if I were to pay enough attention to it's output to alleviate that discomfort it would no longer be vibe-coding.
> you can use the technology of today in ways people with adhd can't.
You say this, but funnily enough I've seriously considered that I might have ADHD and should get examined for it. It feels like there's dozens of symptoms people describe as ADHD that I have, from weird things like sleeping immediately after having caffeine to having trouble listening to a full sentence. My inability to form habits intentionally has been pretty bad for my health already, and I'm not even that old.
reply