HNHacker News
TopNewBestAskShowJobs

Planktonne

2,928 karma · joined April 12, 2019

submissionscomments
Planktonne··on Show HN: Weedout – Safari extension that hides YouTube AI-labeled videos
AI-generated anything sends a message of 'I do not care about quality'. You might feel that that's unfair, but it's the impression you're making on a large segment of your potential audience.
Planktonne··on How accurate have Ed Zitron's AI skeptic predictions been?
I do think you have to consider predictions in the context of the world around them; he's been incorrect, sure, but has he been more incorrect than the predictions of those opposing him? Being less wrong than others is the same as being more right.
Planktonne··on 44% on ARC-AGI-1 in 67 cents
One of these is a much weaker claim than the other.

> yes, LLM's can definitely function in an open world where the rules are fuzzy, changing, undefined, etc.

> At the very least, much better than all GOFAI approaches by far

Planktonne··on EFF to Courts: Don't Rewrite Copyright over AI Hype
The EFF has managed to keep to the letter of its principles while discarding the spirit of them; it's now defending enormous commercial concerns that hoover up and hoard data, at the expense of people creating and sharing information.

A small number of huge corporations keeping tight control of data and access to data is not a pro-freedom stance.

Planktonne··on Dwarf Fortress is getting the mother of all magic updates
I take your point, but those are also all things you can control and manage, albeit indirectly.
Planktonne··on Dwarf Fortress is getting the mother of all magic updates
The point of DF, to a large extent, is precise control and detail. It's the opposite of the approach taken by AI games that want to generalise based on simple prompts.
Planktonne··on Since it was stripped of planetary status, Pluto’s defenders have been fighting
The actual problem here is weirdly aggressive nerds who are incapable of nuance, and insist that an astronomical classification (Pluto is not a planet by the IAU definition) should be used in non-astronomical contexts.

Historically and culturally, Pluto is considered a planet. An IAU classification shouldn't be used to try and retroactively change that, because the IAU does not have control over history and culture. It's a fun fact (did you know that Pluto is not classified as a planet by astronomers?) not a diktat.

Similarly, a banana is 'officially' a berry, which means that it is classified as a berry by botanists. This does not mean it is considered a berry in any other contexts. If you told someone you had made a berry crumble and it was filled with bananas, you would look strange. We have countless examples of this (often focused around food): a tomato does not belong in a fruit salad no matter what botanists say.

Different terms are used with different specificity in different settings. Language and culture and [older disciplines] are not governed by the rules of [newer disciplines or bodies]. If people with 'I love space' t-shirts would stop "um actually"-ing, no one would have a problem.

Planktonne··on [dead]
This is AI-generated without even the faintest effort put into polishing the result.

The site's tagline is

> For collectors & dealers who take books seriously

But they can't even take writing one article seriously; I'm not interested in opinions on books from people who take zero care with their own output.

Planktonne··on Lawmakers added $1 to car insurance policies. That money paid for Flock cameras
It is a useful definition, because it allows you to talk about places where the law might be unjust.

You can't have a functioning democracy if it's impossible to criticise the current setup. This is the whole idea behind newspeak [1]

[1] https://en.wikipedia.org/wiki/Newspeak

Planktonne··on A better SQL in 11 lines of code
This seems harder to read than SQL, and only less verbose if you assume that an SQL database would be built with Prela's limitations in mind, which doesn't feel like a reasonable assumption.
Planktonne··on We found a division by zero bug in FFmpeg with a vibecoded fuzzer
What value are you providing in this scenario?
Planktonne··on Debian polls its developers on AI: permit or ban?
H. G. Wells wrote about how this proverb has limitations [1].

[1] https://en.wikipedia.org/wiki/The_Country_of_the_Blind

Planktonne··on RAG Is Simpler Than You Think
Your brain is incredibly adept at pattern recognition; it doesn't focus on LLM-generated text for the same reason it doesn't stare at wallpaper.

We've all learnt that it's not really communication, and so can be dispensed with.

Planktonne··on Digging the grave of my skills: Hollywood creatives training AI to do their jobs
I disagree; it remains impressive that we can generate video at all, of course, but it's still got the moment-to-moment incoherence that we've seen for years now, and that's before you get to the quality as a whole.
Planktonne··on Digging the grave of my skills: Hollywood creatives training AI to do their jobs
> I have no doubt in my mind that right now [...] an amateur with a high-end last-gen GPU or somebody with maybe a thousand bucks worth of AI credits [...] can crank out a high quality 1h+ feature film with only minor uncanny artifacts.

Then why are there no examples of this? There aren't even examples of steps towards it--the half hour film that's nearly production quality, etc.

Planktonne··on Giving an LLM your prod database is easy. Taking access away is the hard part
Generated prose is devoid of meaningful semantic content.

The one takeaway I got from this article is that deepsql.ai are the last people I would trust to handle this problem, because they can't even explain it themselves.

Planktonne··on I'm becoming AI-blind
You've got it backwards, I think. The people in tech are the ones falling for this endlessly.
Planktonne··on Grand jury declines to indict Ohio man charged with destroying Flock camera
What do you think '86' means? Do you understand that it is not literally the word 'murder'?
Planktonne··on Grand jury declines to indict Ohio man charged with destroying Flock camera
Nonsense.

Using the most motivated possible reasoning, you can just about stretch to an implicit call for murder but

- Definitely not an explicit one

- There are far stronger non-murder interpretations

Planktonne··on AI companies destroy physical books – let's scan rare books before it's too late
A company with no respect for copyright law can't pretend they respect it when it is convenient for them; it's disingenuous, and shows that there is clearly another explanation.
Planktonne··on Copyright does not protect AI-generated content in EU
If 'steal' doesn't fit, then neither do 'eavesdrop' or 'spied'.
Planktonne··on Don't paste the AI, please
> If they use Ai to polish a communication, be it statement, question, etc, and they do it in a thoughtful manner to ensure the final communication clearly and effectively communicates what it should be, there's no problem with that.

They don't though; I have never seen this happen. I have seen many people claim that's what they do, but not convincingly.

Planktonne··on Feature Request: Support AGENTS.md
I really struggle to take this whole thing seriously. ~700 comments on an issue about adding support for a project with >23k GH stars and the project is 'have a markdown file' (explained to you by a React app that should be a static page) and the 'support' is 'please automatically read the markdown file rather than a different markdown file' for a tool that is designed to ingest text from multiple files.

Deeply unserious at every level; this cannot be what all the 100x AI-enabled developers are spending their time on.

Planktonne··on Remote workers report the highest well-being in study of 7,700 employees
I think your comment is an example of why lots of people prefer to work remotely: the 'enforced fun' of it all.

Why is it a problem for you that they don't enjoy eating in canteens? Why does working with you mean signing up to be your best friend?

Planktonne··on Chain-of-Thought Reasoning in the Wild Is Not Always Faithful (2025)
Of course not. Because the article uses the words 'thought' and 'reasoning' and even 'faithful' to mean something other than their normal meanings, but then expects them to behave exactly the same.

Every field has terms of art, and 'reasoning' is one for LLMs. But that doesn't mean it has the same properties as 'reasoning' in other contexts, because you're not referring to the same thing.

Why doesn't my asteroid belt buckle?

Planktonne··on AI Has Plunged the Book Publishing Industry into Utter Chaos
The influx of AI into the creative industries should be considered and responded to as an attack, because it is one. It's malicious actors trying to undermine and destructively extract value.

It's still painfully obvious when a novel was written by AI--the quality is nowhere near good enough--but that's not actually the point; the point is the sheer volume of spam which clogs up search results and harms discovery.

Planktonne··on That Disgraceful, Disreputable, (Wonderful) Form of Punctuation: The Parenthesis
> AI legal hallucinations are a solved computing problem

> those with control of the materials [...] won't share the list with AI companies.

These are definitely claims I would make if I was an AI legal company that hadn't solved hallucinations.

Planktonne··on I used to be excited about new tech, but I rarely am anymore
Echoing others here, but I think part of it is that a lot of the new tech is designed to make things worse.

If I think about new tech, all the examples that immediately spring to mind are negative ones: surveillance cameras, AI-enabled spam & propaganda, autonomous weaponry.

The less flashy advances tend to be harder to get excited about, and they're also often negative. My computers get more expensive, but they also collect more data on me, and it's harder to get one without a bunch of unwanted features.

People used to be excited about new tech because it was exciting and aspirational; I think we've lost a lot of that.

Planktonne··on Israel creates fake think tank in likely attempt to dupe AI chatbots
> He was discussing targeted assassinations. One doesn't "target" an innocent person.

Quite apart from anything else, I don't think these statements hang together, and we wouldn't accept them in any other context. Being targeted is not the same as being guilty.

Planktonne··on AI;DR (AI; Didn't Read)
> people's preference for AI generated content over human content when the provenance of the content is blind, which flips on its head when provenance is not blind

I don't think this shows what you think it shows. AI output (in short clumps) looks like human output; it's designed to do that. Naturally people will sometimes prefer it to other stuff that looks like human output, but crucially this shifts when they realise they're not actually being communicated with.

← PreviousPage 4 of 33Next →