Hacker Newsnew | past | comments | ask | show | jobs | submit | stevula's commentslogin

An LLM can absolutely center a div and then use a browser MCP or playwright to visually validate that it was centered.

Lmao, imagine doing all this instead of just centering the div yourself. This explains why everyone I work with that uses llms is slow af. Its all pseudo-productivity, its kinda similar to how people sit around customizing their nvim config feeling productive but not actually accomplishing anything. The difference is you pay money for this junk.

Imagine being so far behind in your own industry that you think LLMs can only be used for centering divs.

You know what though? It makes sense seeing that you're so aggressively short-sighted that you posted a comment claiming LLMs are useless on a thread about them being critical enough to shape economics and geopolitics in real time.

And by the way, you have no clue just how bad lawyers can be at their own job. If my first lawyer had used an LLM to review his writing, he would have at least caught that he didn't get my name or address right. He also likely would have caught that you having sole custody of your child means you don't pay the other spouse child support. The genius got it backwards.


I think that is how the smarter agents do things? Just like Claude/ChatGPT sometimes does a web search they can do other tool calls instead of just making a statistical guess. Of course it doesn’t always make the bright choice between those options though…


They will also lie and produce output saying it is based on tool execution, without having actually used the tool.

Yes, another layer to cross-check, say, “in kubectl logs I see …” with an actual k8s tool call can help, that is, when the cross-check layer doesn’t lie.

For the time being, IMHO, human validation in key points is the only way to get good results. This is why the tools make experienced people potentially a lot more efficient (they are quick to spot errors/BS) and inexperienced people potentially more dangerous (they’re more prone to trusting the responses, since the tone is usually very professionally sounding).


> it doesn’t always make the bright choice

I'm available for a small fee.


You must be living in absolute opulence :)


You should raise your price


Whereas modern English only distinguishes grammatical number by singular/plural (and Old English had dual), some languages even have trial (three).

Russian distinguishes paucal (few) from plural (many). It’s not super common but there are some other languages that do it.


I thought courts martial and secretaries general (and Knights Templar/Hospitaller, et al) were Anglo-Norman/French borrowings. Do you have any examples of native English phrases following that pattern?


A relative presided over a couple of court martials (1) in the past. Modern usage has largely disconnected it from the past, grammatically (if that is even a thing, except to the true minutaephile)!

(1) https://www.gov.uk/government/publications/court-martial-res...


Whoppers junior


AirPods Pro :)


Passersby


Hards on


Light fantastic


Most tools do, yes.


Probably “more”


There are plenty of unvaccinated people this side of the border and it seems to be a growing trend. If the government is going to intern people in close quarters then they should probably make sure to vaccinate them. I would not be optimistic about the level of medical care sick patients will have access to in this scenario.


If they paid for themselves then the DoT wouldn’t have a multibillion dollar highway budget, and that’s not even including all the state funding.


The claim wasn't they pay for themselves but that they don't generate any income. If we want to look at externalities, we'd also have to figure out how much the Iraq war cost.


That is for Capex. Govt can always easily spend Capex, but Opex has to be covered by the users, whether its roads or trains.


To preserve backwards compatibility and not require all those old sites to update, the legacy behavior would have to be the default, with opt-in for the new behavior.


That is the opposite approach. Also an option. One could also deprecate the call without parameter and force always a parameter with which behaviour. The deprecation could last enough time that those websites would have been rewritten multiple times ;)


The control interface burned into your hardware device will not have been rewritten. And it's not like you can have a flag day where everyone switches over, so the lifespan of those hardware devices isn't that relevant.

Backwards compatibility is a large part of the point of the Web.

You could version everything at whatever granularity you like, but over time that accumulates ("bug 3718938: JS gen24 code incorrectly handles Date math as if it were gen25-34", not to mention libraries that handle some versions but not others and then implicitly pass their expectations around via the objects they create so your dependency resolver has to look at the cross product of versions from all your depencies...)


There is no free lunch. A deprecation warning lasting a decade before erroring will break less that some css boxing models and strict mode in many browsers.


I seem to recall a scene from the book where a man is smoking a cigar in an office and prints out his computer output rather than reading from a screen. It was delightfully retrofuturist (or whatever the opposite of anachronistic is).


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: