Jackson__

Yeah, the blog post took a bit of a weird turn there + the AI slop headers are also quite off putting. I know this may be conspiracy territory, but it does make me wonder whether this is guerilla marketing to ensure big E meets the sale goals for his $1T payment.

Thinking that there is anything worth scraping past the llm-apocalypse is pure hubris imo. It is slop city out there, and unless you have an impossibly perfect classifier to detect it, 99.9% of all the great new "content" you scrape will be AI written.

E: In fact this whole idea is so stupid that I am forced to consider if it is just a DDoS in the original sense. Scrape everything so hard it goes down, just so that your competitors can't.

HN crowd really is not beating the allegations with these replies.

As someone who has spent decidedly too many hours in Aegisubs to create at least somewhat decent subtitles, for projects which have long ago gone the way of the DMCA, these changes feel quite insulting. ASS has been a perfectly cromulent subtitle format and should've been widely adopted.

Though on a related topic, on Netflix it is not just the subtitling that is bad, the translations are awful too. They are cutting every corner so hard I fear they might become a circle.

Can't wait for the Tylenol article on grokipedia to be cited in this court case.

Because GPT-4o is too orange (literally).

I don't think one gets to call these people 'pervert' in the title, then claim that you don't 'yuck their yum'. They yucked their yum before the article even started.

If there are any googlers here, I'd like to report an even more dangerous website. As much as 30-50% of the traffic to it relates to malware or scams, and it has gone unpunished for a very long time.

The address appears to be adsense.google.com.

LLM slop is not just bad—it's degrading our natural language.

> In the late 1950s through the 1970s Herbert Needleman and Clair Cameron Patterson did research trying to prove lead's toxicity to humans. In the 1980s Needleman was falsely accused of scientific misconduct by lead industry associates

> In 1991, together with Claire Ernhart, Scarr was involved as an expert witness on behalf of the lead industry in the lawsuit United States v. Sharon Steel Corp., on the opposite side of Herbert Needleman who was testifying for the U.S. Justice Department owing to his research on the relationship between lead exposure and IQ.[0]

> When the trial was declared open to the public, Scarr initially refused to come and later when she was persuaded she constantly refused to answer questions. Scarr received money from the lead industry for consulting services which creates a conflict of interest.

> Scarr retired to Hawaii in 1997, where she learned scuba diving, even obtaining a rescue diver certification. She also traveled "a lot, especially on cruise ships"

Jesus Christ.

[0] https://en.wikipedia.org/wiki/Sandra_Scarr

A big bet, with none of the mandatory electricity infrastructure figured out. Maybe better to call it a pipedream.

Ubuntu/Linux Mint are actually the worst, for the simple reason that when I tell it to install to a specific, completely empty drive, it will still insert its bootloader onto a completely different drive for absolutely no reason(except it being earlier in the boot order). Absolutely baffling design choice.

All of my cheaper chinese devices without 3.5mm do support it though. It is only this premium tablet that I paid triple what I'd normally spend on which does not work.

Funnily enough, Gemini is also the only one able to read a D20. ChatGPT consistently gets it wrong, and Claude mostly argues it can't read the face of the die that's facing up because it's obstructed (it's not lol).

I just bought a Samsung Galaxy Tab recently. Of course I was well aware of the death of the 3.5mm headphone port, but as I plugged in my 3.5mm to usb-c adapter I was greeted with a new message: "Analog USB C Audio Unsupported.". I tried a pair of regular off-brand usb-c headphones too, which too resulted in the same, with no audio output.

How this is not behavior deserving of some kind of EU fine is a complete mystery to me.

They say they train for ~3 epochs. Could it be that's just not long enough of a training run? I have no idea how many epochs are usually used in those models.

Trying to enact such complete and utter surveillance, as the whole country slowly slides into right-wing populism is an idea that is so utterly ridiculous I struggle to find the right words.

It is beyond stupid, beyond malicious. The closest I might come to describing this would be 'suicidal'.

It fails on any edge case, like all other VLMs. The last time a vision model succeeded at reading analog clocks, a notoriously difficult task, it was revealed they trained on nearly 1 million artificial clock images[0] to make it work. In a similar vein, I have encountered no model that could read for example a D20 correctly.[1]

It could probably identify extra limbs in your pictures if you too made a million example images to train it on, but until then it will keep failing. And of course you'll get to keep making millions more example images for every other issue you run into.

[0] https://huggingface.co/datasets/allenai/pixmo-clocks

[1] https://files.catbox.moe/ocbr35.jpg

This appears to be a classic vision fail on the VLM's part. Which is entirely unsurprising for anyone who has used open VLMs for anything except ""benchmarks"" in the past two god damn years. The field is in a truly embarrassing state, where they pride themselves how it can solve equations off a blackboard, yet couldn't even accurately read a d20 dice roll among many other things. I've tried (and failed) to have VLMs accurately caption images for such a long time, yet anytime I check on the output it is blindingly clear that these models are awful at actually _seeing things_.

Because it is off topic, as the descent of the US into fascism will not impact the tech sphere in any way whatsoever.

Do I even need the sarcasm mark?

I wish open source models would go this route of quality. Instead, every single release since and including flux dev have had some of the worst AI look I've seen so far. Sure these models might produce less mangled bodies, but in terms of actual aesthetics they lack behind even SD1.5 while needing >10x the amount of parameters.

Possibly a fourth group of startup founders and investors that don't want H1B pipelines to dry up as a result of constant bad news.

Holy slop-dogged chatgpt output. Including all the other AI-generated assets on their site, is any human even involved in this product?

Thankfully OAI will fix this, by removing GPT4.1 soon!

> Miyazaki has said over and over that he hates AI image generation

No he has not. He was talking about an AI model that was shown off for crudely animating 3D people in 2016, in a way that he found creepy. If you watch the actual video, you can see the examples that likely set him off here[0].

[0] https://youtu.be/ngZ0K3lWKRc&t=7

>If it were any good I would assume there would be no need to hype it up.

If it were good, there would also be no need for their devs to tell people to temper their expectations, alas... [0]

[0] https://xcancel.com/alexwei_/status/1946477756738629827#m

Mind you, the exact same situation steam faces right now, itch.io had been facing ~3 months ago. If the current state of itch.io is anything to go by, things are still likely to get worse for steam.

Also interesting takeaways from that tweet chain:

>GPT5 soon

>it will not be as good as this secret(?) model

Hey, Summer Memories got an official shout out on their twitter once, someone must have a real weak spot for that one at valve.

Yeah, I find myself disagreeing as well. I think the whole thing also boils down to: If you're self taught and have no achievements, you have no job. So you're not comparing self-taught to academic, but the cream of the crop to the average.