Amazing how someone with your experience can have such poor reading comprehension. He never said anything about permission to use llms. They are saying that to a domain expert like you, you can effectively use LLMs because you will be more likely to spot issues and thus guide it better given your background in the domain. He is saying that some biz bro who know nothing about docusign and esigning who decides one day to use LLMs to create software in that domain is likely to produce working, but buggy and non-compliant software.
As a non expert in a domain, LLMs can seem amazing because you don’t know the field therefore have a harder time spotting errors. As an expert they seem pretty dumb or they’re useful as long as you guide them. It’s akin to Cory Doctorow’s argument about the reverse centaur.
Doesn’t make sense to pay anymore. Ask any LLM to give you a list of FAANG questions asked recently. Then ask “can you restate question number X”. Premium problems locked behind paywall? Just ask it to restate the question and it will.
I didn’t know this until recently either. I grew up on a family where if there was pain, you just take tylenol to deal with it. We always followed the dosage timing and never had issues. So recently when my wife had some chronic pain I suggested taking tylenol but even after two days the pain would not go away and she went to ER. They did a blood check to see if they could find anything. They didn’t find the cause, just suspected food poisoning, but they found her liver enzymes were near dangerous levels. This was even with timing the dosages absed on instructions. Now I am super careful wtih Tylenol.
Not to chastise you or your wife, there is an actual "Liver Warning" on Tylenol Product labels. They warn against 4000mg/day or with alcohol consumption. It's not an on/off switch, the damage occurs on a ramp that achieves a critical point after so much consumption.
Saying this to encourage label reading and to refute those who might try to characterize your wife's experience as anecdotal. And for those who stop reading a page like this one, after the first paragraph that touts "no evidence": https://www.tylenolprofessional.com/safety-and-efficacy/safe... when at the bottom it reveals the Liver Warning.
Alcohol can actually be protective in a paracetamol overdose. Paracetamol itself isn't relatively dangerous, the metabolic product created by the liver is.
Alcohol use acutely reduces the levels of these enzymes which convert paracetamol into it's toxic byproduct as ethanol preferentially binds to the same enzyme for it's own metabolism. Therefore, acute alcohol consumption can reduce the harmful effects of paracetamol.
The reverse is true for overdose in the context of chronic alcohol use, when the liver adapts to long-term alcohol exposure by increasing the levels of enzymes, therefore rapidly metabolising paracetamol into it's toxic byproduct and worsening harm.
Source: doctor
As a side comment, the lack of medical knowledge displayed in HN threads paired with the absolute confidence some verdicts are delivered with is worrying, and something I thought this forum would be better at than other sites (looking at you, reddit). Paracetamol and NSAIDs (including ibuprofens) are safe drugs if taken to the package directions. Some of the ideas shared in this thread about taking one over the other, or not taking one ever etc, are not rational. Please consult your doctor if you are worried about your particular situation, but generally OTC medicines are not dangerous if taken to the package instructions.
Accidental overdoses of paracetamol are exceeding rare, but intentional ones are common.
Its obviosly more dangerous with medicine but I see the same thing with other non-tech topics here like agriculture and construction trades so I always take non-tech topic comments with a grain of salt on here. There is a good diversity of experts on here but of course they can't be expected to correct everybody all the time.
Expert fallacy: I know a lot about X, so I'm also an expert on adjoining topics V, W, Y, and Z. Also on topic "ecks". And "x-rays". And "excerpts" and "excommunication". Ah, hell: I'm smarter than most, so I'm usually right. Especially about things I'm sure I right about.
It appears his wife took under that threshold and only for a few days. 4000mg is roughly 8-10 otc painkillers in a day (depending on the brand).
That said, everyone's physical ability to metabolise drugs is different. I might worry something else was wrong, or at least get a followup test done later, if her liver tox screen showed that, just to rule out another cause.
Everyone's ability to metabolize drugs being different has a lot to do with why the labels end up advising less than is effective for many people. Dextromethorphan surely is NOT ineffective, but at 2 tsp, yea, it aint gonna do much of anything (unless you're CYP2D6 deficient -- then, it may still do a LOT depending on how deficient).
1. NyQuil is not sold here. Not allowed. Probably for that reason! Us northern EUs must be smart, huh?
2. NyQuil is medication. Not what I nor the poster mean. They mean drinking for fun. hmmm whatever to say... Drinking wine, sprit, beer? For the effect of ethanol?
Yup, did some digging on Tylenol/acetaminophen before I had some surgery, and it's a real danger. Oddly enough, it seems downing a whole bottle won't (usually) end your life as it gets mostly directly eliminated rather than metabolized, but taking just around 2-3X the high-strength dose sustained over 3-4 days can destroy your liver — at that point, you have about six weeks unless they find a new liver for you (obviously, there's wide variance in actual dosages and reactions, but that was the general median). For someone with serious acute pain, who doesn't know this hazard, it'd be pretty easy to self-medicate and load up and lose track...
OTOH Ibuprofein, doesn't have anything like that deadly profile, but enough gut and other issues.
Common advice where i am from is to alternate between acetaminaphen/paracetamol and ibuprofen based otc painkillers. One, it reduces your opporunity to overdose, but also some evidence they work better together.
I just having this feeling that these guardrails are there not because it’s super advanced world ending AI. They are there to stop it from doing stupid shit.
> Because it can also happen that CEO/customer will want shiny thing you will deliver the shiny thing and he will have no clue why did you do that, because he forgot that he wanted that - the task has rot away.
Hate this. My boss: “Hey, why is it doing that. Who did this?” You did, you clueless idiot. You asked for it.
Sometimes I seriously am flabbergasted at how many just take what CEOs say at face value. Like, the thought that CEOs need to hype and sell what they’re selling never enters their minds.
Is there any reason to use Claude Code specifically over Codex or Gemini? I’ve found the both Codex and Gemini similar in results, but I never tried Claude because of I keep hearing usage runs out so fast on pro plans and there’s no free trial for the CLI.
I mostly mentioned Claude Code because it's what Mitchell first tried according to his post, and it's what I personally use. From what I hear Codex is pretty comparable; it has a lot of fans. There are definitely some differences and strengths and weaknesses of both the CLIs and the underlying LLMs that others who use more than one tool might want to weight in on, but they're all fairly comparable. (Although, we'll see how the new models released from Anthropic and OpenAI today stack up.) Codex and Gemini CLI are basically Claude Code clones with different LLMs behind them, after all.
IME Gemini is pretty slow in comparison to Claude - but hey, it's super cheap at least.
But that speed makes a pretty significant difference in experience.
If you wait a couple minutes and then give the model a bunch of feedback about what you want done differently, and then have to wait again, it gets annoying fast.
If the feedback loop is much tighter things feel much more engaging. Cursor is also good at this (investigate and plan using slower/pricier models, implement using fast+cheap ones).
Feeling the same. I’m guessing the folks getting good results are literally writing extremely detailed pseudocode by hand?! Like:
Write a class Person who has members (int) age, (string) first name, (string) last name…
But if you can write that detailed…don’t you know the code you want to write and how you should write it? Writing plain pseudo code feels more verbose.
But the AI coding agent can then ask you follow up questions, consider angles you may not have, and generate other artifacts like documentation, data generation and migration scripts, tests, CRUD APIs, all in context. If you can reliably do all that from plain pseudo code, that's way less verbose than having to write out every different representation of the same underlying concept, by hand.
Sure, some of that, like CRUD APIs, you can generate via templates as well. Heck, you can even have the coding agent generate the templates and the code that will process/compile them, or generate the code that generates the templates given a set of parameters.
What happened to Solid Pod? Tim Berners Lee’s project? I feel like it’s another project of this type, and even predates AT, but ever since its announcement hasn’t made much waves.
> I mean, with GUI frameworks it's pretty obvious MS is placing all their chips on WinUI3, even though they are not deprecating any legacy framework.
WinUI3 is dead, lol. I tried to migrate from UWP to WinUI3, but it is literally dead. There doesn’t seem to be any team at MS actively working on it, the community calls have died, and the last build conf didn’t have any WinUI3 talks, all AI stuff. Yes, you can build apps with WinUI3, but development and support for it has stalled and I couldn’t justify moving the companies product over to WinUI3.
As a non expert in a domain, LLMs can seem amazing because you don’t know the field therefore have a harder time spotting errors. As an expert they seem pretty dumb or they’re useful as long as you guide them. It’s akin to Cory Doctorow’s argument about the reverse centaur.
reply