krupan

5 ms·

121.
by krupan 3mo ago
"LLMs are excellent at creating output that looks right but is nevertheless still wrong." "Anthropic has a long history of just lying and making shit up" Every now and then I wonder, if LLMs were created by honest people
122.
by krupan 3mo ago
Thank you for saying this! Every now and then I start to feel like I'm either the only sane person left or I must have gone insane and don't realize it. LLMs amazed us and we've been trying to find a good use for them since
123.
by krupan 3mo ago
Plenty of people telling us not to bother reading code output by LLMs
124.
by krupan 3mo ago
Did it do a Google search to learn all that up to date information?
125.
by krupan 3mo ago
That's not news to me, but this article was a really good reminder of how awful that law is. How have we not gotten it fixed yet??
126.
by krupan 3mo ago
This article seems designed to alarm more than to inform. I'm left with so many questions after reading it. How long does a ship have to sit still before organisms attach to it? How does this anti-attach coating get applied and can it
127.
by krupan 3mo ago
I looked at the ditch use-package post and I didn't see a replacement for my favorite use-package feature: ensure. Did I miss it?
128.
by krupan 3mo ago
The defaults are just fine. The thing you ask for does exist in the form of spacemacs, doom emacs, and whatever other emacs "distributions" exist nowadays (I have never used one) The thing most n00bs should be taught first is tha
129.
by krupan 3mo ago
You haven't noticed how often people here on hacker news lap up LLM company press releases and practically worship them? Defend them vigorously? I think it's totally appropriate to remind everyone how to think about press releas
130.
by krupan 3mo ago
Just FYI, dired without any configuring or customizing is pretty rad, especially with wdired-mode for editing filenames. You don't need to read this blog post to benefit from it. Also, wgrep-mode for editing files is amazing
131.
by krupan 3mo ago
I've been thinking along these lines more and more. Trying to specify exact limits and trying to account for every possible situation when making rules just encourages people to find holes and workarounds. It creates a game, a compet
132.
by krupan 3mo ago
Spoiled with freedom of association and freedom of speech
133.
by krupan 3mo ago
Crazy that we need reminders not to take everything we read in corporate press releases and marketing material at face value
134.
by krupan 3mo ago
Are there any countries NOT on this blacklist?
135.
by krupan 3mo ago
It does because corporations, especially smaller ones competing with bigger ones, actually do get prosecuted for violating restrictions like this a lot more than individuals do.
136.
by krupan 3mo ago
So you'll buy the supplements I'm selling? They'll cure everything that ails you, I promise ;-)
137.
by krupan 3mo ago
It ended with global warming from burning too much coal. I'm not sure that's the comparison you are trying to make though
138.
by krupan 3mo ago
"Feynman’s quote is meant for the author of this post, not people using AI to execute better day to day." Lol! Feynman's quote is in fact for everyone. Refusing to consider that it might apply to yourself is proof of that!
139.
by krupan 3mo ago
I just attempted to verify this myself, but they want my hard earned dollars first. I didn't see any satisfaction guarantee or anything. That already feels like a scam.
140.
by krupan 3mo ago
"Unfortunately, given participant feedback and surveys, we believe that the data from our new experiment gives us an unreliable signal of the current productivity effect of AI tools." That's more than a caveat. That says, th
141.
by krupan 3mo ago
We've been debating how to measure software engineering productivity since long before LLMs. It depends on so many different factors. Scrum is sort of the only tool we have that comes close, but we all hate it. Aside from that, all
142.
by krupan 3mo ago
Hugging Face: Some super smart AI agent hacked us OpenAI: That was us. It was our AI that was smart enough to do this. We even tried to stop it (you know, after we started it), but it outsmarted us. Man, our AI really is super smart. Yo
143.
by krupan 3mo ago
More oversight hinders anyone that is trying to catch up to OpenAI. It's something OpenAI wants
144.
by krupan 3mo ago
Fear is exactly what OpenAI and Anthropic are hoping for. Don't let drive you
145.
by krupan 3mo ago
I mean, once a government starts literally printing money to solve their problems are they not then thoroughly corrupt and dictatorial? You kinda qualified your question into pointlessness ;) Also, the exact process by which governments cr
146.
by krupan 3mo ago
The term "user" maybe made sense before any software was commercialized. It differentiated the programmer of the software from the non-programmer that used the software. Even then it was sometimes used derogatorily (especially w
147.
by krupan 3mo ago
Possibly the author didn't dig this deep into all AI Agent Harnesses, just Open Code?
148.
by krupan 3mo ago
I think you are correct on how it's playing out because AI cannot write software very well all on its own. The dream is that AI is so good that it can write all the code you need from scratch, replacing all and any code written by any
149.
by krupan 3mo ago
"One participant at the retreat noted that permissive licensing was a profound collective mistake, serving as a legal mechanism that enabled the world’s largest corporations to cannibalize volunteer labor..." Agreed. Linux and GN
150.
by krupan 3mo ago
You still have people with very tempting access to a whole lot of data. I don't see how this is different from the current situation other than it's just different (corruptible) people with access.

More ›