tazz.
oz.
+1,341|6816|Sydney | ♥

i use chatgpt a lot- it's like google on steriods. i was able to diagnose a bios issue on my pc that wasn't presenting as a typical bios error by describing symptoms, and it located a reddit post from 5 days ago with two comments. no way i would've found that.

it's great for the troubleshooting side- but when seeking matter-of-fact advice or information- even ~vibe-coding~ the new software developers call it- it needs to be fact checked, frequently.

there's some more interesting implementations in programming specifically- there's a lot of coding that is monotonous,  and some ai tools allow for tab to autocomplete it's guess of what you're about to type out. i love this- and speeds up my work easy 2x.
everything i write is a ramble and should not be taken seriously.... seriously.
uziq
Member
+553|4094
yeah, that works in 90% of the cases but then in 10% of cases, it returns arrant nonsense – worse, advice that can even be harmful. a suggested fix that doesn't solve your tech issue is one thing, but people are increasingly consulting this thing for health or relationship advice – chat-GPT just launched a 'Health' agent ffs! people are going to come to harm because this technology on a quite literal level doesn't know what it's talking about.

whenever you ask it a question on your domain of expertise, you realise quite quickly just how much it is blagging under a thin veneer of the agent's chatty/personable/polite 'voice'. asking it anything doctoral-level, say, on a humanities subject quickly gives away the game. it's regurgitating fine-sounding nonsense, like a jay post in D&ST.

just the other day i asked a very simple query for which it should have been very simple to give a break down in response. i wanted it to compare two speaker models in the same range by the same manufacturer, with only minor changes to specification (mostly size-based). i wanted a breakdown of which speaker is better in which situation. the response from chat-GPT seemed to rely on about 4 reddit threads for the bulk of its sources, and the response it gave to me was garbled nonsense. self-contradictory and sometimes just plain wrong, i.e. talking about one speaker model as if it was the other, and vice versa. of course, as soon as you pull it up and politely query these faults (and politely is the keyword; they actually respond with HIGHER QUALITY replies when you write in a polite tone; not even the company staff know why this happens), it says, "you're right, i got that totally wrong ..."

i totally agree it's like a google on steroids, but it's also occasionally hopelessly shit in a way that google's search algorithm – or at least the old, pre-AI google – isn't/wasn't. sometimes the old way of putting in 15 minutes of cursory reading and self-directed research really does help. for advice on audio hardware, i'm still opting for the 500-page threads on a dusty forum somewhere full of contributions crowd-sourced from a global army of nerds, every single time. and it's not even close. i would rather skimread 20 pages of a thread, even with its noisier signal, than rely on a confident-sounding summary from a chatbot that gets it fucking plain wrong 1 time in 10.

i understand it's going to revolutionise coding and things like the lower rungs of the legal profession. but that's because passing the bar exam or learning to code or whatever is, to a large extent, about memorising the textbooks and official code (whether legal or technical) and regurgitating it back in the right order. of course LLMs are fantastic at that. they've memorised the exam paper backwards and forwards and can recombine it in 10^8 ways.

Last edited by uziq (2026-01-10 05:01:05)

Dilbert_X
The X stands for
+1,841|6747|eXtreme to the maX
Its a chat-bot, it is not an expert system, it does just aggregate information with zero intelligent analysis, and put the information into sentences so it seems like a person.

The few times I've used it its spat out gibberish which almost seems right but isn't.

The most concerning thing to me is that its being used to write code, who knows where that will end up.
Fuck Israel
uziq
Member
+553|4094
it has improved tremendously in recent years/iterations. like if you're asking it day-to-day google query stuff, it will do it.

'give me a trip itinerary for the next week in osaka'. yeah, it'll return like the perfectly simmered and digested breakdown of the ur-tourist experience in osaka. it can do that.

as a day-to-day 'agent' i can see it replacing the culturally ubiquitous habit of 'googling' things. but, yeah, it really doesn't know what the fuck it's doing in any intelligent sense, and the problem is that people are being duped into thinking the 'agent' is somehow an actually intelligent technology that can dispense with advise and expertise, as opposed to presenting extremely concise summaries in 2.5 seconds.

it's what apple's 'hey siri' should have been 10 years ago after they first introduced it. and ironically apple have totally dropped the ball with that particular gizmo.

Last edited by uziq (2026-01-10 04:58:06)

unnamednewbie13
Moderator
+2,095|7413|PNW

uziq wrote:

"you're right, i got that totally wrong ..."
"that was off base of me, and you're totally right to push back on that. i was overreliant on search results from limited sources. if you'd like, we can update with a more grounded comparison between models using real specifications. just let me know!"

slow down and unpack!

Last edited by unnamednewbie13 (2026-01-10 10:32:12)

Dilbert_X
The X stands for
+1,841|6747|eXtreme to the maX

uziq wrote:

as a day-to-day 'agent' i can see it replacing the culturally ubiquitous habit of 'googling' things. but, yeah, it really doesn't know what the fuck it's doing in any intelligent sense, and the problem is that people are being duped into thinking the 'agent' is somehow an actually intelligent technology that can dispense with advise and expertise, as opposed to presenting extremely concise summaries in 2.5 seconds.
From what I read somewhere it basically combines multiple google searches, sifts the information to see what is most popular then combines it in prose so it sounds like a person delivering the result.
It isn't necessarily right about anything as its only measure of rightness is what is on the web including reddit.
Fuck Israel
uziq
Member
+553|4094

Dilbert_X wrote:

uziq wrote:

as a day-to-day 'agent' i can see it replacing the culturally ubiquitous habit of 'googling' things. but, yeah, it really doesn't know what the fuck it's doing in any intelligent sense, and the problem is that people are being duped into thinking the 'agent' is somehow an actually intelligent technology that can dispense with advise and expertise, as opposed to presenting extremely concise summaries in 2.5 seconds.
From what I read somewhere it basically combines multiple google searches, sifts the information to see what is most popular then combines it in prose so it sounds like a person delivering the result.
It isn't necessarily right about anything as its only measure of rightness is what is on the web including reddit.
it's even less executive-functional than that. but, yes, the sales pitch all along has been if only we can feed them all the data in the world, they will somehow escape the chains of their neural network architecture and become all-knowing knowledge machines that can solve physics, cure cancer, and make a bunch of market traders very, very rich, and so on.

the picture is even worse than the fact a huge chunk of the ingested text on the internet is crowdsourced (i.e. uncited and unverifiable) verbiage from the reddit hivemind. the 'expert' knowledge it has ingested has been ripped off wholesale from copyrighted, properly fact-checked and verified sources (like writers and their publishers). it's literally the biggest mass-scale act of piracy and intellectual property theft ever seen – only committed by the likes of Meta, openAI, etc. so they're above the law. recall that one of the original founders of reddit, aaron schwarz, was legally bullied into suicide for ripping a bunch of journal papers using MIT's institutional access ... Meta upscaled this act a hundredfold and nobody blinked an eye.

considering how much americans like to chivvy china for their industrial espionage and copycatting of noble american innovation, it's pretty galling that the most valuable companies on the american stock market just straight up stole their datasets that they train their prized, moated models upon.

further, genuinely good sources for information on the internet, like stack exchange/stack overlfow, have seen their new contributions drop precipitously off a cliff since the LLM era. experts are no longer exchanging their expertise in public. what that means is that, effectively, LLMs are going to be relying on their own outputs to find solutions for anything after ~2022. a big infinite recursion of idiocy. a very big problem when it comes to things like cutting-edge coding or engineering expertise as was frequently found on those websites.

Last edited by uziq (2026-01-22 01:58:40)

Dilbert_X
The X stands for
+1,841|6747|eXtreme to the maX
For a long time my litmus test of a DIY building book was how it instructed on how to use a concrete mixer.
If it said put the aggregate in first then chances are the whole book was copied from other DIY books.
If it said put the water in first then chances were the author had actually used one and the rest of the book might be useful.

Now the test is asking AI how long it takes to get a baby, or if Trump is a paedophile.
Fuck Israel
uziq
Member
+553|4094
sometimes the AI is so unfathomably stupid, like it fails at stuff that an abacus could have ably managed in ancient sumeria. worse, they all seem programmed to provide their responses in such a firm and authoritative tone. i can see the glaring mistakes and omissions really giving a lot of low-IQ normies whiplash in their professional/educational settings.

Board footer

Privacy Policy - © 2026 Jeff Minard