asa123

probably just

consider most naive possible way one would try to deceive users without exorbitant budget -> attempts a few things you naively would do -> recovers it by chance?

if that even is a valid signal though

Hmmm i haven’t had any of these issues. I had one make me some playwright tool and it happily logs me into browsers, saves credentials, scrapes chatgpt for a response/or prompts/inserts file etc. it let me do this in remote browsers as well.

will enter (dummy/worthless) credentials into scripts and even let me commit and push to github but will cry about it every time

i mean even in the US politicans frequently have a JD? Whether you consider that a valid credential in this case for 'law background' is another thing

This is more like a harness for harnesses imo? or a managing interface. some have their own

part of the point is so you don’t need to stick with claude/anthropic apps? you can use other models i’m assuming

I think i built something similar after herdr was released but i can say the reason is at least

tmux is definitely usable but moving between workspaces/repos can be annoying if youre managing mahy many things at once. and difficult to organize. more over having agents manage other agents via tmux can lead to garbled input/output

i havent used codex. claude subagents i think arent the right comparison - the closer comparison would be claude managed agents. but i imagine one reason herdr might be better would be that you’re not stuck in the anthropic ecosystem

the person you are responding to seems to have massively massively downplayed iridium then….

As much as I hear this sentiment, and as much as all the smart software people around me echo the desire for this sentiment to be true; it feels like people say these things, but then at the same time say - well, Claude/Chatgpt is still marginally better for me now. But clearly they are catching up so maybe one day!!!

In that sense, I'm really not sure if Kimi/chinese models are really competing against Anthropic/OAI? Not to mention that for individuals, the subscription model is extremely different, and Kimi subs are (I think) worse value, and chinese API's maybe marginally end up cheaper than using american subscriptions.

i feel like this would be the equivalent of having all CS undergrads program with assembly and assembly only.

it might(?) work? but abstractions exist for a reason (mostly...)

Sometimes I catch myself assuming hacker news is representative of the general population and then I read comments like these

(I am excited to try this out)

aren’t domain specific (large) models already a thing? I feel like they have been for a few years now…

In the end what you said about letting the model know what tools exist is more or less what pi does regarding tools/extensions/plugins.

you can also switch the model at any point too after resuming.

But honestly i dont think there is that big a difference between these “minimal harnesses” to begin with so long as they have the basic functionality…

come on man, it has significantly reduced value, but your comment assumes that if any two people input the same prompt (or sequence of prompts) they would get the same working answer, where its very much possible he had some sequence of guiding, and the thing he is sharing is a working product. which has more value than a prompt that may produce a working product some of the time...

it is quite widely known that the chinese models are just quite shit for math. i think for coding tasks they really do quite alright.

To be clear: it’s not for lack of trying.

i mean marketing in some sense “feels” like fraud when presented in some very specific way, no? at some level it’s: generate interest about a thing when there didn’t exist any interest before. sometimes they generate interest about something that has no substance and that feels kind of fraudulent (this isn’t that case)

but in this case, this is product extremely boosted for super viral marketing, and this is the first time i’ve read about this boost since hearing of the product whereas everybody only really talks of its substance. feels like a magician kept a trick going for a very long time, people forgot they were watching a magic show, then the curtains come down and they reveal the deceit after (i’m being slightly melodramatic)

you’ve got to be kidding me. i know it ISNT fraud but i can’t help but feel like this SMELLS like fraud

I think you're overthinking it. you clearly care a lot about the output and respecting the reader. you're not the problem lol

what do you mean:

"what are your observations on feeling as if sharing that output though?"

and "Should I try to share the actual transcript of it as well, I really wish if some good ethics and internet ettiquette could be established about it."

could you rephrase the question

for STT, it's literally you saying it, with a model transcribing, and then another model correcting a little, and you also have control over editing it. I don't think any of the arguments on "etiquette re: sharing AI output" apply here.

this was guest posted onto his blog i believe so you may want to hold that thought

this is like an extreme extreme version of how the general regards letter from the UN “denouncing” said “horrible” activity engendered by nation “X”

with no teeth, stuff like this starts to feel a little funny+sad

i think math is close to art (just to be a contrarian, but kind of really)

its somewhat funny that math people are in a conundrum as to support or not support but this might partially be because some wish to believe that math itself is and can be useful and therefore accelerating is good

but the art people have no such delusions so they’re just strictly against

imo proof writing is more akin to art than coding/tech but…

not op but firecracker exists, as does docker sbx (not container!),

firecracker is a lot more of a headache to setup than docker sbx (not container!) but if you can get it going it probably “feels” the best

of a different variety some people feel better using stuff like bubblewrap/fire jail but idk if these are still microvm as opposed to the above

but it’s my opinion (perhaps completely criticizable), that sandboxing for personal home use machines is 1) somewhat overkill 2) somewhat theatre 3) psychologically sometimes exhausting and unrealistic always assuming the worst is going to happen and 4) not really worth the time and creates quite a bit of friction that there wasn’t previously. In the end I spent more time tinkering with the sandbox to get it “just” right that it drained my time actually using the agents so …

beware (s)he who open hegel for pleasure

it really isn’t

clear would be something like

piss-cheap - it’s-alright-i guess - okay-relax - ouch-my-wallet

it's really not

looking from the tail end of the generation, many of my friends ( and also I ) were led to believe somehow that things like marijuana, psychedelics, non-Alcohol have such low consequences that are entirely a rounding error compared to alcohol. which is ...

i also would like to know an answer to this

more than anything it's valid enough as a frame of reference for what many people want to do. if they went and invented a new term for it i'd reckon a lot of people just wouldn't even click on it

more recently, I've (allegedly) seen a arxiv preprint by a proper tenure track professor that was suspiciously AI slop (and incorrect nonetheless!)

I feel like I'm losing it

My problem was more that it seemed to start with one thesis, and then throughout the article took long detours to market other achievements and then managed to not say very much at the end...