andy99

27 ms·

391.
by andy99 10mo ago
I don’t really find it better when someone ads a disclaimer. What am I supposed to do then? There’s still an expected default behavior of reading it, and if I don’t I need to confront them and say “I don’t care what you got an LLM to say, w
392.
by andy99 10mo ago
The article is interesting and I hadn’t thought about this really, but it’s pretty obvious that the underlying reason is that personal injury lawyers are completely undifferentiated, and so can only compete on being recognized. Real estate
393.
by andy99 10mo ago
> has broad (and cross-border) value I’m wondering if this is the primary use case and that’s why I don’t really get it? I see the online aspect, but it comes with a whole other set of problems that for most cases it’s hard to imagine ar
394.
by andy99 10mo ago
How would gift cards circumvent this? Dollar for dollar I can only imagine a transaction getting more attention if done in gift cards, certainly one over 10k probably isn’t even possible
395.
by andy99 10mo ago
What is the core argument why a gift card would ever be used instead of cash? I can see if someone didn’t have any money and had a card and wanted to try and sell or exchange it, otherwise? Cash might have some limitations but none that are
396.
by andy99 10mo ago
What red flags does cash throw up that are circumvented by gift cards?
397.
by andy99 10mo ago
I’ve had these, they would have got the CEO from LinkedIn and my number is googleable from having resumes online. Someone is doing this reconnaissance on different companies.
398.
by andy99 10mo ago
It’s famous for being an allegory isn’t it? Isn’t this like saying Animal Farm remains popular even though we’ve proven that animals don’t actually self organize like in the book?
399.
by andy99 10mo ago
You’re definitely on to something, people wouldn’t criticize as much as they are otherwise, they’d ignore it. I think the whole point is that in optimizing for instruction following and boring realism we’ve lost what could have been some un
400.
by andy99 10mo ago
Well, birthdays are merely symbolic of how another year's gone by and how little we've grown. No matter how desperate we are that someday a better self will emerge, with each flicker of the candles on the cake we know it's no
401.
by andy99 10mo ago
to me something like for key in possible_keys: if key in collection: ... is fine and isn’t subject to your disadvantage.
402.
by andy99 10mo ago
There was a discussion the other day where someone asked Claude to improve a code base 200x https://news.ycombinator.com/item?id=46197930
403.
by andy99 10mo ago
The chinchilla paper says the “optimal” training data set size is about 20x the number of parameters (in tokens), see table 3: https://arxiv.org/pdf/2203.15556 Here they do 80B tokens for a 4B model.
404.
by andy99 10mo ago
I wonder if they know about this, basically training on LLM output can transmit information or characteristics not explicitly included https://alignment.anthropic.com/2025/subliminal-learning/ I’m curious, they ha
405.
by andy99 10mo ago
Very cool, I’m probably thinking too much but why are they seemingly hyping this now (I’ve seen a bunch of this recently) with no M5 Max/Ultra machines in sight. Is it because their release is imminent (I have heard Q1 2026) or is it t
406.
by andy99 10mo ago
I’d like to know how they chat-tuned it. Getting the base model is one thing, did they also make a bunch of conversations for SFT and if so how was it done? We develop chatbots while minimizing interference with the normative judgments
407.
by andy99 10mo ago
Yeah it’s trained to do that somewhere though it’s not necessary malicious. For RLHF (the model fine tuning) the HF stands for human feedback but is really another trained model that’s trained to score replies the way a human would. And so
408.
by andy99 10mo ago
Only if leadership lets them. Right now (anecdotally) a lot of “leaders” don’t understand the difference between AI generated and human generated work, and just look at loc as productivity so all incentives are on AI coding, but that will c
409.
by andy99 10mo ago
I think the problem is in what “proven” means. People that don’t know any better will just do that all with LLMs and still deliver the giant untested PRs but with some LLM written tests attached. I vibe code a lot of stuff for myself, mostl
410.
by andy99 10mo ago
If there is regulation, it should be about monopolies in general and not trying to micromanage hiring. Companies behave this way because they are in a position to do so. In a real competitive environment they wouldn’t. Poorly thought throug
411.
by andy99 10mo ago
Is this actually different from just guaranteeing some metrics? Like if you have a document processing “agent” that extracts fields from forms, you’d have an accuracy threshold and have some checks set up to verify this? Does “outcome billi
412.
by andy99 10mo ago
What do you mean about not doing evals? Just literally that you don’t run any benchmarks or do you have something against them?
413.
by andy99 10mo ago
At this point if I was openAI I wouldn’t bother fixing this to give pedants something to get excited about.
414.
by andy99 10mo ago
Maybe a stupid question, how do you verify the verification program? If an LLM is writing it too, isn’t it turtles all the way down, especially with the propensity of AI to modify tests so they pass?
415.
by andy99 10mo ago
https://news.ycombinator.com/item?id=46197930 They changed the name making it very hard to find
416.
by andy99 10mo ago
I’ve been using LLMs to code for some time and I look at it differently. I ask myself if I need to understand the code, and if the answer is yes I don’t use an LLM. It’s not a matter of discipline, it’s a sober view of what the minimal amou
417.
by andy99 10mo ago
Autonomy generally, not just privacy. You never know what the future will bring, AI will be enshittified and so will hubs like huggingface. It’s useful to have an off grid solution that isn’t subject to VCs wanting to see their capital retu
418.
by andy99 10mo ago
Yes this has conservative psy-op written all over it /s
419.
by andy99 10mo ago
This has nothing to do with whether the technology is valuable or not, it’s about cramming superficial treatment of trendy topics into academic degree rewuirements, which whatever one thinks of AI should be frowned upon.
420.
by andy99 10mo ago
That was my thought, it feels like something a career college or high school would do. Are CS students going to have to take a “how to talk to chat gpt course”? That’s probably less condescending than making an arts student or someone else

More ›