HNHacker News
TopNewBestAskShowJobs

paulluuk

930 karma · joined June 13, 2019

submissionscomments
paulluuk··on Research-Driven Agents: When an agent reads before it codes
This sounds like it would work, but honestly if you've already read all 30 papers fully, what do you still need to llm to do for you? Just the boilerplate?
paulluuk··on Just Put It on a Map
This is great, and it also feels like a great way to answer the question "Where should I buy a house if I want to be close to the center but not in the expensive area?".

> Let’s play a guessing game. How much more valuable is land in Manhattan than in the Bronx? Take a guess, then scroll down for the answer.

As someone who has never been in New York and doesn't live in the US, I knew beforehand that I would fail this test very hard, haha.

paulluuk··on 3D-Knitting: The Ultimate Guide
I've tried, but I've been to so many thrift stores and never found anything I liked or which fit well. I'm a tall guy and the men's section of thrift stores is just filled with what I assume is the clothes that widows donate after their husbands die. And I don't mean that as a joke, half of it looks like something a small boxy train conductor in the 50s would wear.
paulluuk··on Willingness to look stupid
> One meeting did in his promo.

Although true, I feel it's worth adding here that the problem is that PM. While looking stupid by asking questions can "do you in" when working with incompetent managers like that, I'd argue that most managers will look at results -- and asking dumb questions can lead to much better results compared to just staying quiet and hoping for the best.

paulluuk··on 3D-Knitting: The Ultimate Guide
I bought a 200 dollar jacket and it had holes in it within months, just from regular use. I have an old 3 dollar shirt I bought years ago and it's only now beginning to show wear.

One problem this shows, is that as a consumer I have no idea what the hell is quality clothing. Clearly, expensive does not always mean high quality. And I'm not buying "brand" clothing either.

paulluuk··on Tell HN: I'm 60 years old. Claude Code has re-ignited a passion
I understand your sentiment. I personally would never use a textbook for anything code related, if there's no proper documentation online then I wouldn't touch it with a ten-foot pole, haha.

However, even though I've never worked with CassandraDB, I feel pretty confident that I could do it with Claude Code. Not just "do it for me", but more like "I have done a lot of database migrations in my time, but haven't worked with CassandraDB in particular. Can you explain to me the complexities of this migration, and come up with a plan for doing it, given the specifics of this project?"

That question alone is already a massive improvement over a few years ago. I don't feel like I was using my "critical thinking muscles" when I tried to figure out how the hell to get hadoop to run on windows, that was just an exercise in frustration as none of the documentation matched the actual experience I was getting. Doing it together with Claude Code would be so much easier, because it'll say something like "Oh yeah this is because you still need to install XYZ, you can do that by running this line here: ...".

Now I'm not saying that Claude Code, and agentic in general, isn't taking away some of my critical thinking: it really is. But it also allows me to learn new skills much more quickly. It feels more like pair programming with someone who is a better programmer than me, but a much worse architect. The trick is to keep challenging yourself to take an active role in the process and not just tell it to "do it", I think.

paulluuk··on Can you reverse engineer our neural network?
Don't we already harvest more food than humans could ever eat, and have a huge pharmaceutical industry? I get what you're saying but these two examples seem counterproductive imho.

Which begs the question: what would actually be a good field to apply human potential towards? I agree that finance, sales and ads are very low on that list.

paulluuk··on Games Look Bad: HDR and Tone Mapping (2017)
I feel like this is very much a personal preference thing. They even called out Horizon Zero Dawn for looking very bad, and Zelda for looking very good.. while in my opinion the exact opposite is true.
paulluuk··on NIH limits scientists to six applications per year
The question is: is AI breaking the system, or was it always broken and does AI merely show what is broken about it?

I'm not a scientist/researcher myself, but from what I hear from friends who are, the whole "industry" (which is really what it is) is riddled with corruption, politics, broken systems and lack of actual scientific interest.

paulluuk··on An average human breathes out roughly 1kg of carbon dioxide a day
21 lbs -> 9.5 kg For those not working for the empire ;)
paulluuk··on Why is AI so slow to spread?
I work in R&D, and although I haven't signed an NDA, I think it's best if I don't elaborate too much. But basically we have a large dataset of shows and movies for which we already know how well they did with specific audiences, but we didn't know why exactly. So we use LLMs to reverse-engineer a large amount of metadata about these shows, and then use traditional ML to train a model that learns which feature appeal to which audiences.

Most stuff is obvious: nobody needs to tell you what segment of society is drawn to soap operas or action movies, for example. But there's plenty of room for nuance in some areas.

This doesn't guarantee that it actually becomes a succesful movie or show, though. That's a different project and frankly, a lot harder. Things like which actors, which writers, which directors, which studio are involved, and how much budget the show has.. it feels more like Moneyball but with more intangible variables.

paulluuk··on Why is AI so slow to spread?
I know this is just a casual comment, but this is a genuine concern I have every day. However, I've been working for 10 years now and working in music/video streaming has been the most "societal value" I've had thus far.

I've worked at Apple, in finance, in consumer goods.. everywhere is just terrible. Music/Video streaming has been the closest thing I could find to actually being valuable, or at least not making the world worse.

I'd love to work at an NGO or something, but I'm honestly not that eager to lose 70% of my salary to do so. And I can't work in pure research because I don't have a PhD.

What industry do you work in, if you don't mind me asking?

paulluuk··on Why is AI so slow to spread?
It really depends on the use-case. I currently work in the video streaming industry, and my team has been building production-quality code for 2 years now. Here are some things that are going really well:

* Determine what is happening in a scene/video * Translating subtitles to very specific local slang * Summarizing scripts * Estimating how well a new show will do with a given audience * Filling gaps in the metadata provided by publishers, such as genres, topics, themes * Finding the most "viral" or "interesting" moments in a video (combo of LLM and "traditional" ML)

There's much more, but I think the general trend here is not "chatbots" or "fixing code", it's automating stuff that we used armies of people to do. And as we progress, we find that we can do better than humans at a fraction of the cost.

paulluuk··on Bypassing Google's big anti-adblock update
Interesting, I also just installed Firefox because of OPs comment, and I'm amazed at how much faster it is then Chrome.
paulluuk··on Apple vs the Law
I don't understand what you feel is "protectionist" about this? I would say that the US pressuring the EU on behalf of big corporations is arguably "protectionist", but I don't think that's what you mean.

But even if policies make companies less "capable" and less "competitive": that completely ignores what effect they have on society. I bet that a company that was given a free pass to use slavery would be very capable and very competitive -- but is that what we want for our society?

paulluuk··on I deleted my second brain
I was replying specifically to the statement that writing verbosely is a form of "skilled writing", which I don't agree with. Simply being verbose does not make your writing any better.
paulluuk··on I deleted my second brain
> "I would have written a shorter letter, but I did not have the time." - Blaise Pascal

It takes much more skill to write concise than verbose.

paulluuk··on AI Improves at Improving Itself Using an Evolutionary Trick
It's really a choice: do you want to waste compute or do you want to waste potential?

While prioritizing higher scorers for selecting progenitors will initially mitigate some of the problems, you will eventually end up with hundreds of thousands of agents that only learned to repeat the letter "a" a million times in a row, which is a huge waste of processing.

paulluuk··on AI Improves at Improving Itself Using an Evolutionary Trick
I'm not sure if it's ridiculous if you factor in something like copilot. Heck, even just your IDE's built-in autocomplete (which only finishes the current variable name) can get close to being responsible for 20% of your code, with tools like copilot I think you can even more easily hit that target.
paulluuk··on Snorting the AGI with Claude Code
Me laughing as a human non-frontend dev having to do anything related to CSS

The number of times that my manager or coworkers have rejected proposals for technical solutions because I can't make a webpage look halfway decent is too damn high.

paulluuk··on What methylene blue can (and can’t) do for the brain
Maybe this is a very european sentiment, but I "trust" an italian or german pharma lab just as much as a dutch or american one.
paulluuk··on Seedance 1.0
I work on AI solutions for a major video streaming company, and the problem with VEO3 is that it doesn't have any consistency between prompts. E.g. I can not upload a reference image of what a character looks like, and if I say in one video "the old priest bends down" and in the next video "the old priests picks up the coin", the priest will look very different between shots.

Veo3 does support image to video, so what you can do is create an image that is the start of a scene, and then use that to generate the actual scene. Unfortunately, Veo3 is really bad at this. I expect this will improve over time.

Although I'm not super excited about this Seedance model personally, I do really like that it focuses on consistency between shots. I hope this puts pressure on increased performance from Veo3 in that regard.

paulluuk··on What methylene blue can (and can’t) do for the brain
I order plenty of things online (From other countries) and other than having to wait a week, it's trivially easy. Is it that much different when it's medicinal products?
paulluuk··on The rise of judgement over technical skill
You've seen good managers promoted to engineers? ;) I have seen this happening, usually the engineers with the best technical AND people skills are first made lead developer, and eventually "team lead". After team lead they can climb the corporate ladder with titles like "junior vice president" or "senior director".
paulluuk··on The ‘white-collar bloodbath’ is all part of the AI hype machine
On a side note, I do worry about the energy consumption of AI. I'll admit that, like the silicon valley tech bros, there is a part of me that hopes that AI will allow researchers to invent a solution to that -- something like fusion or switching to quantum-computing AI models or whatever. But if that doesn't happen, it's probably the biggest problem related to AI. More so even than alignment, perhaps.
paulluuk··on The ‘white-collar bloodbath’ is all part of the AI hype machine
Around the time when bitcoin started to get serious public attention, late 2017, I remember feeling super hyped about it and yet everyone told me that money spent on bitcoin was wasted money. I really believed that bitcoin, or at least cryptocurrency as a whole, would fundamentally change how banking and currencies would work. Now, almost 10 years later, I would say that it did not live up to my believe that it would "fundamentally" change currencies and banking. It made some minor changes, sure, but if it weren't for the value of bitcoin, it would still be a nerdy topic about as well known as perlin noise. Although I did make quite a lot of money from it, though I sold out way too soon.

As a research engineer in the field of AI, I am again getting this feeling. People keep doubting that AI will have any kind of impact, and I'm absolutely certain that it will. A few years ago people said "AI art is terrible" and "LLMs are just autocomplete" or the famous "AI is just if-else". By now it should be pretty obvious to everyone in the tech community that AI, and LLMs in particular, are extremely useful and already have a huge impact on tech.

Is it going to fulfill all the promises made by billionaire tech CEOs? No, of course not, at least not on the time scale that they're projecting. But they are incredibly useful tools that can enhance efficiency of almost any job that involves setting behind a computer. Even just something like copilot autocomplete or talking with an LLM about a refactor you're planning, is often incredibly useful. And the amount of "intelligence" that you can get from a model that can actually run on your laptop is also getting much better very quickly.

The way I see it, either the AI hype will end up like cryptocurrency: forever a part of our world, but never quite lived up to it's promises, but I made a lot of money in the meantime. Or the AI hype will live up to it's promises, but likely over a much longer period of time, and we'll have to test whether we can live with that. Personally I'm all for a fully automated luxury communism model for government, but I don't see that happening in the "better dead than red" US. It might become reality in Europe though, who knows.

paulluuk··on Trying to teach in the age of the AI homework machine
Yeah, I've had teachers like that, who tell you that you're a "waste of life" and "what are you doing here?" and "you're dumb", so motivational.

I guess this "tough love" attitude helps for some people? But I think mostly it's just that people think it works for _other_ people, but rarely people think that this works when applied to themselves.

Like, imagine the school administration walking up to this teacher and saying "hey dum dum, you're failing too many students and the time you've spent teaching them is a waste of life."

Many teachers seem to think that students go to school/university because they're genuinely interested in motivated. But more often then not, they're there because of societal pressure, because they know they need a degree to have any kind of decent living standard, and because their parents told them to. Yeah you can call them names, call them lazy or whatever, but that's kinda like pointing at poor people and saying they should invest more.

paulluuk··on Spaced repetition systems have gotten better
> They try, write bad flashcards, don't see results, and give up.

But if that's the case, then wouldn't a program that takes a long text (like a book) and creates GOOD flashcards, be way superior over someone making their own bad flashcards?

paulluuk··on LegoGPT: Generating Physically Stable and Buildable Lego
Indeed, I would be very curious to see how their robots would actually build that sofa. Although the robots aren't really part of the model of course, they're just a little extra.
paulluuk··on Xiaomi MiMo Reasoning Model
If you put it like that, I absolutely agree with you, except that I suppose I don't really consider this an exchange of knowledge but more like the release of an open-source project: the only thing they need to publish are instructions on how to use it. I don't think they’re really interested in anyone improving their model by themselves or reproducing the work. It would be amazing if they did, though!
← PreviousPage 2 of 11Next →