The same happens with whisper-large-v3 on Chinese transcription: silence is transcribed to something like "please upvote, share and favourite this video". I suspect they trained the model on some random YouTube video without carefully picking really useful data.
In Chinese, it always added something like "For study/research purpose only. Please delete after 48 hours." This is what those volunteers added in subtitles of (pirated) movies/shows.
There is so much damning evidence that AI companies have committed absolutely shocking amounts of piracy, yet nothing is being done.
It only highlights how the world really works. If you have money you get to do whatever the fuck you want. If you're just a normal person you get to spend years in jail or worse.
There's actually a lot of court activity on this topic, but the law moves slowly and is reluctant to issue injunctions where harm is not obvious.
It's more that the law about "one guy decides to pirate twelve movies to watch them at home and share with his buddies" is already well-settled, but the law about "a company pirates 10,000,000 pieces to use as training data for an AI model (a practice that the law already says is legal in an academic setting, i.e. universities do this all the time and nobody bats an eye)" is more complicated and requires additional trials to resolve. And no, even though the right answer may be self-evident to you or me, it's not settled law, and if the force of law is applied poorly suddenly what the universities are doing runs afoul of it and basically nobody wants that outcome.
What’s ironic to me is that had these companies pirated only a single work, wouldn’t that be a chargeable crime?
Clearly Bonnie and Clyde shouldn’t have been prosecuted. Imagine they were just robbing banks for literary research purposes. They could have then used the learnings to write a book and sell it commercially…
Or imagine one cracks 10000 copyrighted DVDs and then sells 30 second clips… (a derived work).
To me, for profit companies and universities have a huge difference — the latter is not seeking to directly commercially profit from copyrighted data.
There is a distinction that must be made that very few people do, but thankfully the courts seems to grasp:
Training on copyright is a separate claim than skirting payment for copyright.
Which pretty much boils down to: "If they put it out there for everyone to see, it's probably OK to train on it, if they put it behind a paywall and you don't pay, the training part doesn't matter, it's a violation."
So if I download copyrighted material like the new disney movie with fansubs and watch it for training purposes instead of enjoyment purposes it's fine? In that case I've just been training myself, your honor. No, no, I'm not enjoying these TV shows.
Because it's important to grasp the scale of these copyright violations:
* They downloaded, and admitted to using, Anna's Archive: Millions of books and papers, most of which are paywalled but they pirated it instead
* They acquired Movies and TV shows and used unofficial subtitles distributed by websites such as OpenSubtitles, which are typically used for pirated media. Official releases such as DVDs tend to have official subtitles that don't sign off with "For study/research purpose only. Please delete after 48 hours" or "Subtitles by %some_username%"
OpenSubtitles is almost exclusively used with pirated media. Official copies come with official subtitles. OpenSubtitles itself is legal, but that's not the point at all.
Whether it’s legal slash fair use to train on copyrighted material is only one of the questions currently being asked though. There’s a separate issue at play where these companies are pirating the material for the training process.
By comparison, someone here brought up that it might be transformative fair use to write a play heavily based on Blood Meridian, but you still need to buy a copy of the book. It would still be infringement to pirate the e-book for your writing process, even if the end result was legal.
If they would buy material at a large scale, the seller might require them to sign a contract that requires royalty if the material is used for training an AI. So buying legally is a way to put yourself into a trap.
The only thing I've been able to find is the note that since copyright is federal law, state contract law actually can't supersede it, to wit: if you try to put a clause in the contract that says the contract is void if I use your work to make transformative fair-use works (or I owe you a fee), that clause is functionally unenforceable (for the same reason that I don't owe you a fee if I make transformative fair-use works of your creations in general).
The dead corpses of filmmakers and authors and actors are buried in unmarked graves out behind those companies' corporate headquarters. Unimaginable horror, that piracy. Why has no one intervened?
>If you're just a normal person you get to spend years in jail or worse.
Not that I'm a big fan of the criminalization of copyright infringement in the United States, but who has ever spent years in jail for this?
Besides, if it really bothered you, then we might not see this weird tone-switch from one sentence to the next, where you seem to think that piracy is shocking and "something should be done" and then "it's not good tht someone should spend time in jail for it". What gives?
EDIT: apparently he wasn't in jail, he was on bail while the case was ongoing - but the shortest plea deal would still have had him in jail for 6 months, and the penalty was 35 to 50 years.
> Besides, if it really bothered you, then we might not see this weird tone-switch from one sentence to the next, where you seem to think that piracy is shocking and "something should be done" and then "it's not good tht someone should spend time in jail for it". What gives?
What a weirdly condescending way to interpret my post. My point boils down to: Either prosecute copyright infringement or don't. The current status quo of individuals getting their lives ruined while companies get to make billions is disgusting.
> Either prosecute copyright infringement or don't
This is the absolute core of the issue. Technical people see law as code, where context can be disregarded and all that matters is specifying the outputs for a given set of inputs.
But law doesn’t work that way, and it should not work that way. Context matters, and it needs to.
If you go down the road of “the law is the law and billion dollar companies working on product should be treated the same as individual consumers”, it follows that individuals should do SEC filings (“either require 10q’s or don’t!”), and surgeons should be jailed (“either prosecute cutting people with knives or don’t!”).
There is a lot to dislike about AI companies, and while I believe that training models is transformative, I don’t believe that maintaining libraries of pirated content is OK just because it’s an ingredient to training.
But insisting that individual piracy to enjoy entertainment without paying must be treated exactly the same as datasets for model training is the absolute weakest possible argument here. The law is not that reductive.
That's not the same as piracy though. He wasn't downloading millions of scientific papers from libgen or sci-hub, he was downloading them directly from jstor. Indeed, none of his charge was for copyright infringement. It was for stuff like "breaking and entering" and "unauthorized access to a computer network".
No, they couldn't, since the then-novel and untested strained interpretation of the CFAA that the prosecutor was relying on has since been tested in the courts and soundly rejected.
I haven’t seen any accusations that they’ve done that, though. Usually people get pirated material from sources that intentionally share pirated material.
> Hitting URLs denied by robots.txt has been argued to be just that.
"Has been argued" -- sure, but never successfully; in fact, in HiQ v. LinkedIn, the 9th Circuit ruled (twice, both before and on remand again after and applying the Supreme Court ruling in Van Buren v. US) against a cease and desist on top of robots.txt to stop accessing data on a public website constituting "without authorization" under the CFAA.
Part of the accusation comes from the fact that Swartz accessed the downloads through a MIT network closet, which AI companies wasn't doing. The equivalent to that would be if openai broke into a wiring closet at Disneyland to download Disney movies.
The CFAA is vague enough to punish unauthorized access to a computer system. I don't have an example case in mind, but people have gotten in trouble for scraping websites before while ignoring e.g. robots.txt
The CFAA might be vague, but the case law on scraping pretty much has been resolved to "it's pretty much legal except in very limited circumstances". It's regrettable that less resourced defendants were harassed before large corporations were able to secure such rulings, but the rulings that allowed scraping occurred before AI companies' scraping was done, so it's unclear why AI companies in particular should be getting flak here.
Aaron Swartz was not jailed or even charged for copyright infringement. The discussion and the comment I replied to is centered around US companies and jurisdiction.
There could be a moral question. For example a researcher might not want to download a pirated paper and cause loss to a fellow researcher. But it becomes pretty stupid to pay when everyone, including large reputable companies endorsed by the government, is just downloading the content for free. Maybe his research will help developing faster chips to win against China, why should he pay?
Would it be a "fair use" to download pirated papers for research instead of buying?
Also I was gradually migrating from obtaining software from questionable sources to open source software, thinking that this is going out of trend and nobody torrents apps anymore, but it seems I was wrong?
Or another example: if someone wants to make contributions to Wine but needs a Windows for developing the patch, what would be the right choice, buy it or download a free copy from questionable source?
Researchers don't get paid when their papers are downloaded, though. They pay to have their papers downloaded, and the middleman makes money on both sides. Piracy is the only moral option for them. There is a reason every single competent professor in the western world will email you a free copy of their papers if you ask nicely.
No, if you revolutionize both the practice and philosophy of computing and advance mankind to the next stage of its own intellectual evolution, you get to do whatever the fuck you want.
I get the common cynical response to new tech, and the reasons for it.
We wish we lived in a world where change was reliably positive for our lives. Often changes are sold that way, but they rarely are.
But when new things introduce dramatic capabilities that former things couldn't match (every chatbot before LLMs), it is as clear of an objective technological advance as has ever happened.
--
Not every technical advance reliably or immediately makes society better.
But whether or when technology improves the human condition is far more likely to be a function of human choices than the bare technology. Outcomes are strongly dependent on the trajectories of who has a technology, when they do, and how they use it. And what would be the realistic (not wished for) outcome of not having or using it.
For instance, even something as corrosive as social media, as it is today, could have existed in strongly constructive forms instead. If society viewed private surveillance, unpermissioned collation across third parties, and weaponizing of dossiers via personalized manipulation of media, increased ad impact and addictive-type responses, as ALL being violations of human rights to privacy and freedom from coercion or manipulation. And worth legally banning.
Ergo, if we want tech to more reliably improve lives, we need to ban obviously perverse human/corporate behaviors and conflicts of interest.
(Not just shade tech. Which despite being a pervasive response, doesn't seem to improve anything.)
Well, wait, if somebody writes a computer program that answers 5 of 6 IMO questions/proofs correctly, and you don't consider it an "advance," what would qualify?
Either both AI teams cheated, in which case there's nothing to worry about, or they didn't, in which case you've set a pretty high bar. Where is that bar, exactly? What exactly does it take to justify blowing off copyright law in the larger interest of progress? (I have my own answers to that question, including equitable access to the resulting models regardless of how impressive their performance might be, but am curious to hear yours.)
The technology is capable in a way that never existed before. We haven't yet begun to see the impacts of that. I don't think it will be a good for humanity.
Social networks as they exist today represent technology that didn't exist decades ago. I wouldn't call it an "advancement" though. I think social media is terrible for humans in aggregate.
I notice you've motte-and-baileyed from "revolutionize both the practice and philosophy of computing and advance mankind to the next stage of its own intellectual evolution" to simply "is considered an 'advance'".
You may have meant to reply to someone else. recursive is the one who questioned whether an advance had really been made, and I just asked for clarification (which they provided).
I'm pretty bullish on ML progress in general, but I'm finding it harder every day to disagree with recursive's take on social media.
Except that the jury’s (at best) still out on whether the influence of LLMs and similarly tech on knowledge workers is actually a net good, since it might stunt our ability to critically think and problem solve while confidently spewing hallucinations at random while model alignment is unregulated, haphazard, and (again at best) more of an art than a science.
Well, if it's no big deal, you and the other copyright maximalists who have popped out of the woodwork lately have nothing to worry about, at least in the long run. Right?
It's not about copyright _maximalism,_ it's about having _literally any regard for copyright_ and enforcing the law in a proportionate way regardless of who's breaking the laws.
Everyone I know has stories about their ISP sending nastygrams threatening legal action over torrenting, but now that corporations (whose US legal personhood appears to matter only when it benefits them) are doing it as part of the development of a commercial product that they expect to charge people for, that's fine?
And in any case, my argument had nothing to do with copyright (though I do hate the hypocrisy of the situation), and whether or not it's "nothing to worry about" in the long run, it seems like it'll cause a lot of harm before the benefits are felt in society at large. Whatever purported benefits actually come of this, we'll have to deal with:
- Even more mass layoffs that use LLMs as justification (not just in software, either). These are people's livelihoods; we're coming off of several nearly-consecutive "once-in-a-generation" financial crises, a growing affordability crisis in much of the developed world, and stagnating wages. Many people will be hit very hard by layoffs.
- A seniority crisis as companies increasingly try to replace entry-level jobs with LLMs, meaning that people in a crucial learning stage of their jobs will have to either replace much of the learning curve for their domain with the learning curve of using LLMs (which is dubiously a good thing), or face unemployment, and leaving industries to deal with the aging-out of their talent pools
- We've already been heading towards something of an information apocalypse, but now it seems more real than ever, and the industry's response seems to broadly be "let's make the lying machines lie even more convincingly"
- The financial viability of these products seems... questionable right now, at best, and given that the people running the show are opening up data centres in some of the most expensive energy markets around (and in the US's case, one that uniquely disincentivizes the development of affordable clean energy), I'm not sure that anyone's really interested in a path to financial sustainability for this tech
- The environmental impact of these projects is getting to be significant. It's not as bad as Bitcoin mining yet, AFAIK, but if we keep on, it'll get there.
- Recent reports show that the LLM industry is starting to take up a significant slice of the US economy, and that's never a good sign for an industry that seems to be backed by so much speculation rather than real-world profitability. This is how market crashes happen.
They can. I don't think anyone got prosecuted for using an illegal streaming site or downloading from sci-hub, for instance. What people do get sued for is seeding, which counts as distribution. If anything AI companies are getting prosecuted more aggressively than "ordinary people", presumably because of their scale. In a recent lawsuit Anthropic won on the part about AI training on books, but lost on the part where they used pirated books.
But in that case even though filming isn't technically distribution, it's clearly a step to distributing copies? To take this to the extreme, suppose you ripped a blu-ray, made a thousand copies, but haven't packaged or sold them yet. If the FBI busted in, you'd probably be prosecuted for "conspiracy to commit copyright infringement" at the very least.
You seem to equate "training" (with scare quotes) with someone actually pirating a blu-ray, but they really aren't equivalent. Courts so far have ruled that training is fair use and it's not hard to see why. Unlike copying a movie almost verbatim (as with ripping a blu-ray), AI companies are actually producing something transformative in the form of AI models. You don't have to like AI models, or the AI companies' business models, but it strains credulity to pretend ripping a blu-ray is somehow equivalent to training an AI model.
Who's to say why I downloaded and am now watching a movie? Is it for my enjoyment? Is it because I'm training my brain? How is me training my brain any different from companies training their LLMs?
Same goes for recording: I'm just training my skills of recording. Or maybe I'm just recording it so I can rewatch it later, for training purposes, of course.
>Who's to say why I downloaded and am now watching a movie? Is it for my enjoyment? Is it because I'm training my brain? How is me training my brain any different from companies training their LLMs?
None of this is relevant because Anthropic was only left off the hook for training, and not for pirating the books itself. So far as the court cases are playing out, there doesn't appear to be a special piracy exemption for AI companies.
>Same goes for recording: I'm just training my skills of recording. Or maybe I'm just recording it so I can rewatch it later, for training purposes, of course.
You can certainly use that as a defense. That's why we have judges, otherwise there's going to be some smartass caught with 1KG of coke and claiming it's for "personal consumption" rather than distribution.
None of this matters in reality, though. If you're caught with AV gear in a movie theater once, you'd likely be ejected and banned from the establishment/chain, not have the FBI/MPAA go after you for piracy. If you come again, you'd likely be prosecuted for trespassing. In the cases where they're going after someone in particular for making these rips, they usually have a dossier of evidence, like surveillance/transaction history showing that the same individual has been repeatedly recording movies, and watermarks correlating the screenings that the person has been in to files showing up on torrent sites.
> If you're caught with AV gear in a movie theater once, you'd likely be ejected and banned from the establishment/chain, not have the FBI/MPAA go after you for piracy
Good example, because this is exactly what websites are doing with LLM companies, who are doing their damnest to evade the blocks. Which brings us back around to "trespassing" or the CFAA or whatever.
IANAL, but reading a bit on this topic: the relevant part of the copyright law for AI isn't academia, it's transformative work. The AI created by training on copyrighted material transforms the material so much that it is no longer the original protected work (collage and sampling are the analogous transformations in the visual-arts and music industries).
As for actually gathering the copyrighted material: I believe the jury hasn't even been empaneled for that yet (in the OpenAI case), but the latest ruling from the court is that copyright may have been violated in the creation of their training corpus.
That is not the case here - I never encountered this with whisper-large-v3 or similar ASR models. Part of the reason, I guess, is that those subs are burnt into the movie, which makes them hard to extract. Standalone subs need the corresponding video resource to match the audio and text. So nothing is better than YouTube videos which are already aligned.
At least for English, those "fansubs" aren't typically burnt into the movie*, but ride along in the video container (MP4/MKV) as subtitle streams. They can typically be extracted as SRT files (plain text with sentence level timestamps).
*Although it used to be more common for AVI files in the olden days.
I have also found them inside mkvs as the subtitle track. I think SRT was the default because most content was ripped from DVD/BD, but now most of the content is from streaming sources and you need to convert the subtitles anyway.
Indeed, with another model I would get persistent transcriptions of silent parts into 'Thanks for watching!' or '[MUSIC]'. Pretty dumb that this failure mode wasn't caught in some QA process, and there are now multiple transcription models suffering from the same issue. Having silent parts in your input audio seems like it should be a very common occurrence...
When I was taught mathematics, the zero value was always considered the most important edge case. You prove something for N=0 (or N=1), then for N=M+1.
It's even more important in audio DSP: processing near-zeroes can end up being extremely CPU intensive, look up denormal/subnormal floats.
Yeah, I studied mathematics (algebra and number theory) and zero is the point, often sporting discontinuities, or weird asymptotic behavior.
Quite a lot of algorithms use some form of division and zero is the only number in our typical structures (Z, Q, R, C), that cannot be used to divide with.
In machine integer arithmetics, one must also beware division by -1, which can convert MIN_INT into MIN_INT with a signed overflow and violate some arithmetics invariants, such as sign (negative divided by negative is _usually_ positive).
Makes total sense, execution time is bounded. The point is it's still a case you must consider (what if near-zero is distinct from zero and significant?)
Considering that if you DO use VAD (voice activity detection), it's the best open weights voice recognition model by a very wide margin, it's quite good. I'd be willing to be that commercial products that "don't have this problem" are using VAD as well, and that this is well known to them. But Whisper is just the weights, and I suppose a simple reference implementation, not a full product.
> What good is a speech recognition tool that literally hears imaginary voices?
Well, if it is supposed to work after silence detection, then it is good for speech recognition I guess. It's like blaming a wheel why is it circular, you can't sit on it. It's a part of a larger machine.
On the other hand, I can imagine that when things get quiet and the signal-to-noise ratio gets close to zero, random background audio (or randomness introduced in the transcription model) will be enough to tickle a critical number of neurons and elicit hallucinations.
The related thought exercise is this: Try scanning across the band with an AM or sideband radio, and after a while your brain will start to wonder "was that a voice I just heard, or music perhaps?" when in reality it was just environmental static.
Yes, you are holding it wrong. The good of it is that it does not output imaginary voices when used with VAD.
Show us a technology with better results that does not use VAD. If you can’t, then I’m not sure what you’re arguing against except superficialities so inconsequential that I can’t comprehend the condescension. The results speak for itself
So if a tool has a process to have it perform at its best then it's a problem?
Do you also moan that before applying glue to a surface or it won't stick? Or if you need to drill a guiding hole before making a larger one in wood? Or that you need to use truly prime numbers for a security key to actually be safe?
How is it not the case? It is unusable without VAD or editing. I don't understand what you're questioning
I agree their products could be better "end to end" integrated. Meanwhile there is a continuously-improving field of work for detecting speech (which Whisper is incapable of). They offer official "cookbooks" with guidance on an approach they recommend: https://cookbook.openai.com/examples/whisper_processing_guid...
> At times, files with long silences at the beginning can cause Whisper to transcribe the audio incorrectly. We'll use Pydub to detect and trim the silence.
You put in 2+2 - the right figures. The machine says 4 - the right answer. If you put in the wrong figures, like 3+3, will the machine still say 4? It's easy to make a machine that always says 4.
The people who asked him that question, however, probably got a different scam demonstrated to them every every. Remember the Mechanical Turk? Babbage's reply paints him very honestly. It shows that he couldn't even conceive that someone might try to trick the royal court (or whoever it was) into accepting a fake device.
When YouTube began building automatic transcriptions for captions, it regularly flagged any noise or music -- typically industrial noise -- with "[foreign]"
If it couldn't understand it, it was "foreign" for the longest time.
Haha, yes, it's fair when English subtitles write something like [speaks Japanese], especially when at least one of the characters is not supposed to understand what's being said (when they do, it's more appropriate to write "[in Japanese]: let's go shopping!").
Netflix sometimes takes the cake with what I consider the most outrageous option: writing "[in English]" when they mean "in whatever language the protagonist considers native", which is mind-bogglingly wrong and hilarious at the same time.
They do this with the English subtitles of the German production "Die Kaiserin" ("The Empress"): whenever Sisi is speaking in another language, say French, the subtitles will say "[in French] I love you...", and when she switches back to German they will say "[in English] I love you...". WTF, Netflix? Note this is unrelated to understanding German; it's mostly Netflix looking down on its customers and assuming they cannot comprehend there are people in the world for whom their native tongue is different to the viewer's native tongue.
This has happened in more shows, enough to know it's not a fluke, though Netflix is inconsistent about it.
Can confirm as well, although to my recollection it just shows up as if it's a word the transcription model heard, not "[foreign]" in brackets like with "[Music]" or "[Applause]". It's especially weird to me because I recall the auto-transcriptions being reasonably serviceable when they first rolled them out, only to degrade over time to the point where it was hallucinating the word "foreign" and dropping letters from words or using weird abbreviations (like "koby" for "kilobyte", "TBTE" for "terabyte", or, most memorably weirdly, transcribing the phrase "nanosecond-by-nanosecond" as "nond by nanc") if it didn't decide it heard another one entirely.
I also noticed a couple of months ago that YouTube seems to have quietly rolled out a new auto-transcription model that can make reasonable guesses at where capitalization, punctuation, and sentence boundaries should go. It seems to have degraded even more rapidly than the old one, falling victim to the same kinds of transcription errors. Although the new one has a different hallucination in silence and noise that it wasn't able to classify (which, incidentally, its ability to recognize things like music and applause seems worse than the old one's): where the old model would have hallucinated the word "foreign", the new one thinks it's hearing the word "heat", often repeated ("Heat. Heat.").
That's interesting, the few times I tried playing with whisper, I had the impression that YouTube style videos or random cellphone videos was something it did particularly bad with (compared to movies). My guess at the time was that most of the training material might be sub titles and raw screen plays.
The videos I tried to transcribe were also Mandarin Chinese, using whisper-large-v3. Besides the usual complaints that it would phonetically "mishear" things and generate nonsense, it was still surprisingly good, compared to other software I played around with.
That said, it would often invent names for the speakers and prefix their lines, or randomly switch between simplified and traditional Chinese. For the videos I tested, intermittent silence would often result in repeating the last line several times, or occasionally, it would insert direction cues (in English for some reason). I've never seen credits or anything like that.
In one video I transcribed, somebody had a cold and was sniffling. Whisper decided the person was crying (transcribed as "* crying *", a cough was turned into "* door closing *"). It then transcribed the next line as something quite unfriendly. It didn't do that anymore after I cut the sniffling out (but then the output switched back to traditional Chinese again).
Similar in the English model. Pretty clear they trained on YouTube videos where creators will put that in otherwise silent sections to ensure it shows up for people with CC on.
This reminds me, some years ago as Google was expanding its translation service, someone tried translating text into and out of an obscure African language (don't recall which) and it always came out as weird Biblical-sounding semi-gibberish.
My revelation was that machine translation needs a corpus of bilingual documents to learn from, and if the language is sufficiently obscure, there may not be any bilingual documents except for the Bible, which missionaries have translated into just about every language on Earth.
This is totally happening with other models too, at least with Spanish. Many transcriptions will end with something that roughly translates to "Thanks for watching!" even if it's never present in the original audio.
Right, maybe my definition of overfitting was wrong, I always understood it more as trying to optimize for a specific benchmark / use case, and then it starts failing in other areas.
But the way you phrase it, it’s just “the model is not properly able to generalize”, ie it doesn’t understand the concept of silence also makes sense.
But couldn’t you then argue that any type of mistake / unknown could be explained as “overfitting” ? Where do you draw the line ?
I don't think so. Overfitting = the model was too closely aligned to the training data and can't generalize towards *unseen* data. I think it saw "silence" before, so it's not overfitting but just garbage in, garbage out.
> [By] that definition any incorrect answer can be explained by “overfitting to training data”.
No it doesn't, for instance some errors would be caused by under fitting. The data could also be correct but your hyperparameters (such as the learning rate or dropout rate) could cause your model to overfit.
> Where do you draw the line between “overfitting to training data” and “incorrect data” ?
There's no need to draw a line between two explanations that aren't mutually exclusive. They can (as in this case) both be true. Overfitting is the symptom; dirty data is the cause.
Silence is never put in the subtitles of a film, since it isn't necessary. The viewers can tell that nothing is being said if there are actors on the screen. And in situations where there are no actors, then there will be a subtitle to indicate what is going on, like "[rock music plays]".
Subtitle authors use this silence to fit in meta information and have done so since the closed captions era.
Proper data cleaning procedures would be to strip this meta data from any subtitle sources. Since this wasn't done, this is fundamentally a classification issue. It may also be an over-fitting issue, but that is secondary to the classification problem.
I think it's a data quality problem first, which might lead to a sort of overfitting as a consequence.
How would the AI know that a series of zero-amplitude audio samples should generate the string "[silence]"?
It can only know that if the vast majority of silent audio segments in the trainser are consistently labelled with that string. But that doesn't seem to be the case: Silence is either not labeled at all, or labeled with all kinds of different markers or labeled with unrelated things, like copyright credits.
So even if the model successfully learns a generalized representation of the concept of "silence", it's not clear at all which of all the different labels it should use for that concept.
So what might happen is that the model then starts to overfit on the tiny variations of the individual silence segments, in a desperate attempt to devise some kind of system behind the all the different "silence" labels - which will of course go wrong spectacularly as such a system doesn't exist. (Or if it does, is entirely accidental and not something that should be learned)
It's actually because it is incapable of recognising when it does not know the answer. It will give you the nearest match, even if that is completely incorrect.
And the German is “subtitles of [public broadcaster] for [content network], 2017
I'm not sure this is really overfitting, the network does exactly what the training data demands. According to the training data silence art the end transcribes to a copyright notice or subtitle credits
Overfitting would be replicating overly specific details. Like if a specific pattern of silence (or quiet noise) matched to specific copyright notices.
But in this case the behavior seems to generalize over multiple languages, with the model choosing representative "outro silence" captions depending on the language. Which is consistent with the training data showing that outro silence is captioned.
If the model was generalizing perfectly it would show something like "[subtitle credits here]" but that'd be demanding a bit much.
Transcribing outro silence as silence despite the training data consistently transcribing outro silence differently from regular silence would be underfitting
The optimizer is functioning correctly, and the pattern really exists in the training data. But consider:
- This behavior damages the model's performance on out of sample data; every word you predict during silence increases the transcript's Word Error Rate.
- These translation credits are an artifact of our training data, and not a reflection of the process we are modeling (spoken language).
So, while you are correct about the mechanism at work here, it is still correct to call learning a spurious pattern which damages our performance "overfitting".
No. Because there would have been indtances in the data where silence was labelled correctly. But the model couldnt handle the null case, so it over fit on the outros. But generally it fit on the random error in the label of the null feature. Which is what overfitting is
Exactly. Underfitting would be if the model doesn't pick up on the fact that outro silence is labeled differently from regular silence and transcribes them the same
Side-note: it's also yet more evidence that AI companies hoover all data with no regard for legality or copyright status, the very same offences that got other people in jail or with heavy fines.
Which ... would be overfitting. It picks up on a pattern in the training data that we don't want it to pick up on and which causes it to generalize poorly.
How is it overfitting if the data is garbage in the first place? Saying it's overfitting in this context has no meaning as there is no alternative that maximizes the utility function we're training for?
As I didn't see one correct definition of overfitting:
overfitting means that the model is too closely aligned to the test data, picked up noise and does not generalize well to *new, unseen* data. think students that learn to reproduce questions and their answers for a test instead of learning concepts and to transfer knowledge to new questions that include the same concepts.
while this sounds like overfitting, I'd just say it's garbage in, garbage out; wrong classification. the training data is shit and didn't have (enough) correct examples to learn from.
That's the magic of money. Download your favorite artist's discography for personal use? If the MPAA had its way (and it occasionally has), torrenting that could bankrupt you.
The AI industry - soaking up every bit of media available online for commercial purposes, often reproducing it nearly identically - has enough money and capital to influence things its way. And only its way, in case anyone was hoping this might change anything at all for the little guy.
> Download your favorite artist's discography for personal use? If the MPAA had its way (and it occasionally has), torrenting that could bankrupt you.
I don't think that there are any clear examples of cases where ONLY downloading has resulted in huge fines. All the big bankrupting level fines have been for both downloading and sharing.
You mention that 'torrenting' could bankrupt you, and that is true, but the main reason for the huge fines are that you are taking part in distribution rather than just 'downloading for personal use'.
> I don't think that there are any clear examples of cases where ONLY downloading has resulted in huge fines.
They [1, and others] been hunting and fining downloaders for over a decade now, with the only "evidence" being IP addresses connected with the torrent [2].
>with the only "evidence" being IP addresses connected with the torrent [2].
Is that an unreasonable assumption? As much as people like to come up with excuses like "I had open wifi!" or "I was running a TOR node", judges don't seem inclined to believe them, probably for the same reason they don't seem inclined to believe excuses like "somebody took my car on a joyride and then returned it!" for parking tickets. Remember, both non-commercial copyright infringement lawsuits and parking tickets are tried in civil court, which means the standard is "preponderance of evidence", not "beyond reasonable doubt".
You are missing the point I was replying to, specifically that parent suggested people were only hunted for creating/uploading pirated content, not merely participating in the torrent.
That’s a really interesting distinction. Clearly there’s an “original crime”, the first person to rip the CD and put it online (or whatever kids do to pirate music nowadays).
But then if I download a file, create a copy, and share it with you, have I done anything wrong?
To all intents and purposes, seeding is an act of reproduction. You, while keeping your copy, create copies of (parts) of the file and share it to someone else to allow them to assemble a new, second copy.
Whether this is, or should be, a crime is a different question altogether. The main point I was making is that it’s the copying/sharing to other people which seems to be a crucial element in these prosecutions.
That’s likely intentional: the last thing the *AA folks want is a decision that creating a copy of a copyrighted work for your own personal use is not a crime. But it does seem the courts have decided: making a copy for someone else is indeed illegal.
But both are illegal? I suspect if it came out that some torrent seeder was actually part of some sort of piracy ring responsible for ripping the movies, they'd get far stiffer penalties than the few thousand $ fine that typical torrenters get. Moreover isn't AI companies also "keeping content available"?
The whole point of the thread is that AI companies are getting away with piracy but individuals aren't. But the reality is that AI companies aren't getting away with it (a judge ruled that Anthropic must face trial over their use of pirated books).
More specific to this thread is that claim that "ONLY downloading" hasn't resulted in fines for anyone. So far as I can tell, this is true. People are just quibbling over how someone who's torrenting somehow counts as "only downloading", even though their client is uploading.
Yes, but torrenting is not ONLY downloading, it's both. The articles you link are very clearly talking about 'Sharing' (from link 2: "File sharing consists of both download and upload of a file.").
Given the lack of sense in treating each peer as a lost sale for damages, I think we can safely say they're only interested in making examples out of people and would absolutely go after people for only downloading if the law permitted. Thankfully it's not, but maybe they lobby to make changes in that direction to try and curb future AI industry shenanigans.
You contradict yourself. There were numerous public cases where they chased people downloading few mp3s just for themselves, and made into example case with massive fines.
If you don't understand how torrents work on technical level I suggest at least some shallow reading. Property rights holders don't care about details, as long as you tick the box of sending a single packet to somebody, off to court with ya.
> There were numerous public cases where they chased people downloading few mp3s just for themselves
If this is true, I have been unable to find any. Can you please share? In all of the cases I was able to find, the huge fines were based on also uploading.
> If you don't understand how torrents work on technical level I suggest at least some shallow reading
This is a bit patronising, and I'm not sure what point you're trying to make. My point is that the only prosecutions I've been able to find are where they were able to prove uploading as well as downloading (and yes, the fact that someone used BitTorrent makes it a slam-dunk, because the protocol makes it impossible to download without also uploading). Are you trying to argue that someone who torrents a copyrighted work doesn't also share it?
The fight about digitized media for personal (entertainment / informational) use were the early aughts. The precedents crafted then don't immediately translate to these cases (novel transformative work from protected materials), and the new precedents have to account for the fact that universities have been training via "piracy" for ages.
(The magic of money factors in to the extent that they can afford the lawyers to remind the court that this isn't settled law yet).
The fact that this is propping up the entire AI industry adds additional weight. When legislating or deciding court cases, some won't be willing to pop the cash cow, some will be worried about falling behind countries that don't enforce copyright evenly. IP owners are trying to go after the AI industry, with only mixed to poor success.
Hard to justify that they can't afford to pay when they have multi-billion dollar valuations and are apparently paying hundreds of millions to get a single engineer.
Anthropic is going to trial over pirating books for training. The judge was pretty clear that even if training is fair use, the training material must be obtained legally.
These regurgitations combined with proof that a model is familiar with a work could be sufficient evidence to force discovery to determine if the work was pirated.
What's insane is copyright. How come you can own intellectual property but not pay a property tax? The ecosystem would be much healthier if to get copyright protections you should declare value of your IP (that you are obligated to sell for if the buyer pops up) and pay tax on this for every year you hold the IP.
> if to get copyright protections you should declare value of your IP (that you are obligated to sell for if the buyer pops up) and pay tax on this for every year you hold the IP
I think this would have some unpalatable consequences. Let's say an author is writing a modestly successful book series: it's not going to make them rich, but it's commercially viable and they care a lot about it for its own sake. Under this system, if the author declares a value commensurate with the (quite small) pure economic value of the IP, they have to live in fear of their right to continue working on their creation being abruptly taken away from them at any point. If they instead declare a value commensurate with the economic value + the extra value that it has to them personally, the resulting tax liability could easily tip the balance and destroy their ability to pursue their writing as a career.
You are always free to update the value before paying tax. If somebody is willing to pay more than it's worth to you they probably have an idea how to turn it into more economic value for the society. So the society should allow them to do that. For a price, of the tax. What I'm proposing is about the financial rights. Individual right, like the right to call yourself author of any given creation should be inalienable.
There are always some cases on the edge. The question is if saving them is worth the cost of the major players running rampant.
>What's insane is copyright. How come you can own intellectual property but not pay a property tax? The
Most jurisdictions that have "property tax" only apply it on certain types of property, most commonly real estate. So it's not that weird that IP isn't taxed.
Can you imagine if we evaluated property taxes this way? Yeah, nice single family home, better hope nobody offers you the same amount you paid for it or it's back to apartment living for you and your kids.
And it seems to be because the training data is largely unofficial subtitles from movies. Which often have a string like "Translated by X" at the end of the movie which is often silent while credits roll.
Looks like they used more official sources for German - there, silence is apparently hallucinated as "Untertitelung des ZDF für funk, 2017" according to one of the comments on the issue. Which makes sense, as the public broadcasters' "Mediathek" is probably the largest freely available resource of subtitled videos in Germany. I wonder if the ZDF gave its approval for it being used for LLM training though?
I'm being made to pay for Autobahnen I barely use, finance kindergartens despite not having a child, and made to pay into public pensions with little hope of getting close to the same value out. All under threat of imprisonment, many without a way to even refuse (not that I'd want to) The only thing that sets the pubic broadcasting fee apart is that it's collected separately from taxes in an attempt to reduce the influence politicians have on broadcasters
This person refers to the German television and radio fee (Rundfunkgebühren).[1] It is a state-mandated system that ensures free (as in free speech) and (relatively) neutral public broadcasting institutions. There is a constant and engaged discussion, because every household in Germany has to pay this fee. Exceptions are made only for low-income households.
A constant discussion, lately fueled by extremist parties (AfD) who feel treated unfairly by (amongst others) the public broadcasters (which has parallels to Trump's recent campaign against public broadcasters in the US).
Can't argue them - Tageschau always has been trashtalking people with the wrong opinion.
Back in 2011, Tageeschau openly rallied against Muslims and wanting public broadcasting gone was a leftist position. The whole thing is completely asinine to anyone who remembers.
Ah, ok, thanks for the info, TIL! "We are funk – the first public service content network that started on October 1, 2016. We create online-only content on social networks and third-party platforms, including YouTube, Instagram, Snapchat, TikTok, Spotify, Apple Music or Twitch for 14-29 year-olds." (https://presse.funk.net/das-ist-funk/, scroll down for the English version). I live in Germany, and I even watch public broadcasters regularly, but this is the first time I have heard about funk (I even initially thought it was misspelled, usually it's written with a capital F). But I'm not part of the targeted audience (not now, nor even back in 2016 when it was launched), so all good...
> We have a public service mandate, which means that we have very clear responsibilities according to the state media treaty. For us, this means that our top priority is actually reaching our target audience, namely approximately 15 million people living in Germany between the age of 14 and 29 who have internet access
It's not a binding contract for sure but I don't think that OpenAI or other AI scraper is their target.
Let's not forget that some of real pirates (for example, corsairs) also were legal and performed legitimate pirate activities to ships of foreign countries.
>Let's not forget that some of real pirates (for example, corsairs) also were legal and performed legitimate pirate activities to ships of foreign countries.
In other words there are activities that are legal or not depending on whether you have authorization from the state. That describes many things. For instance you synthesize meth without a license from the DEA/FDA, you're a "drug cartel" or whatever. But if you do it with a license you're a "pharmaceutical company", and you're not making "meth", you're making "desoxyn".
Families of fentanyl overdose victims would disagree. Moreover it's not hard to find examples of "legal if the government authorizes it" for killings. Cops and soldiers, for instance.
It takes quite a leap of logic to blame someone ingesting a toxic quantity of a substance on the person who manufactured the substance. When someone drinks bleach do we blame the company that makes the bleach?
I suspect privateers would have been offended at being called pirates, but is this what is going on? If it specifically a Chinese AI company pirating Hollywood for example sure, but it seems it's more of a everyone firing at everyone situation.
How is this evidence of that fact? Honest question.
I can see how this might show that subtitles from online sub communities are used, or that maybe even original subtitles from e.g. DVDs are used. But isn't it already known and admitted (and allowed?) that AI uses all sorts of copyrighted material to train models?
> Indeed, the captioning is copyrighted work and you are not legally allowed to copy and redistribute it.
Unless you qualify for one of the many exceptions, such as fair use
Training isn’t recreating or distributing so copyright won’t apply if the ruling is actually consistent with the intention of the law, which it may not.
Using copyrighted materials and then meaningfully transforming it isn’t infringement. LLMs only recreate original work in the same way I am when I wrote the first sentence of this paragraph because it probably exists word for word somewhere else too
It’s been determined by the judge in the Meta case that training on the material is fair use. The suit in that case is ongoing to determine the extent of the copyright damages from downloading the material. I would not be surprised if there is an appeal to the fair use ruling but that hasn’t happened yet, as far as I know. Just saying that there is good reason for them to think it’s been allowed because it kind of has; that can be reversed but it happened.
There's no reason to think those cases will go any differently. As far as I know, the ruling would have to be appealed at this point. I am only commenting to say that there is reason to think this is true:
> But isn't it already known and admitted (and allowed?)
You seemed to be confused about why this person believed that:
> No, and I don't see where you got that from.
And I wrote a comment intended to dispel your confusion. The above commenter thought that it was allowed because a judge said it was allowed; that can be appealed but that's the reason someone thinks it's allowed.
> There's no reason to think those cases will go any differently. As far as I know, the ruling would have to be appealed at this point.
Trial court rulings aren't binding precedent even on the same court in different cases, so its quite possible that different cases at the trial level can reach different conclusions on fair use on fairly similar facts, given the lack of appellate precedent directly on point with AI training.
A single verdict about a specific case (13 authors vs META) does not mean it's legal for companies to steal IP from other companies which has evidently been going on for some years now.
Those other companies have lawyers powerful enough to change jurisdiction in many countries in order to "protect their IP".
The Chinese subtitles for silence use a common mark for pirated media in that language, according to other commentors here. In general it's pretty likely that if you're finding non professional subtitles they were distributed with pirated media in some form, that's where you get the most fan subs after all
> were distributed with pirated media in some form,
I disagree with this conclusion. I've used e.g. the opensubtitles dataset for some data-analysis in the past. It's a huge dataset, freely available and precisely intended for such use. Now, if all the data in the opensubtitles dataset is legal, is another point.
So one might argue that using this opensubtitles dataset, makes one complicit to the illegal activities of opensubtitles themselves, IDK: IANAL.
The contention is that the specific translated text appears largely from illegal translations (i.e., fansubs) and not from authorized translations. And from a legal perspective, that would basically mean there's no way they could legally have appropriated that material.
> But isn't it already known and admitted (and allowed?) that AI uses all sorts of copyrighted material to train models?
Technically, everything is copyrighted. But your question is really about permission. Some of the known corpuses for AI training include known pirate materials (e.g., libgen), but it's not known whether or not the AI companies are filtering out those materials from training. There's a large clutch of cases ongoing right now about whether or not AI training is fair use or not, and the ones that have resolved at this point have done so on technical grounds rather than answering the question at stake.
Whisper is unusable IMO because of the hallucinations. Widely documented. Removing silence from audio clips helps, but even then it will auto correct grammar, translating bilingual speech, etc. Improved in the latest audio models but not solved [1]
I wouldn't describe it as "unusable" so much as needing to understand its constraints and how to work around them. I built a business on top of Whisper [1] and one of the early key insights was to implement a good voice activity detection (VAD) model in order to reduce Whisper's hallucinations on silence.
Thanks for noticing. It took a lot of effort to optimize the pipeline every step of the way. VAD, inference server, hardware optimization, etc. But nothing that would compromise on quality. The audio is currently transcribed in its original speed. I'll be sure to publish something if I manage to speed it up without incurring any losses to the WER.
That's the problem with raws large models, it should always be coupled with satellite small models and logic. It's (probably) easier to detect hallucinations using a traditional ML/DL model that can catch mismatches (it's easy to build a synthetic dataset for this) than transcribing. And the simplest piece of code can detect a silence and that it should match no text.
In Russian it often hallucinates "Субтитры сделал DimaTorzok" ("Subtitles by DimaTorzok") at the end of things. Interestingly, I wasn't able to find any YouTube videos with that name in the subtitles, so it's not like it's in a lot of training data.
I tried googling this and found questions from Telegram users why voice messages recognition sometimes produces this phrase and who is this person. Also I found this thread [1] claiming that the subtitles by DimaTorzok are coming from some Russian youtube videos on gaming like [2].
"In the future, everyone will be world-famous for 15 minutes" _in a microniche techno-linguistic community, at a time and choosing of the swirling AI clouds_
TL;DR: Whisper occasionally hallucinates and credits “Nicolai Winther” at the ends of Norwegian transcriptions during silent audio segments, likely because the real Nicolai Winther - a former YouTuber who created subtitles—appears frequently in its (likely YouTube‑based) training data. This highlights how limited Norwegian training (only ~266 hours) can cause the model to overfit on specific names and phrases when uncertain.
In the comments to that Github issue, by alentodorov:
in romanian, i’ve noticed multiple instances where the transcripts ends with “nu uitati sa da-ti like si subscribe” which, as you might easily infer , translates to “don’t forget to like and subscribe”.
I wonder if hallucinated copyright claims (esp. like the ZDF one at the bottom of the OP) will be introduced as evidence in one of the court cases against "big AI"
"Big AI" is transparent and open about the fact they use all sorts of copyrighted material to train the data. How would "we see an exact chunk of text from our copyrighted material" add to that?
It appears they have not been training on the official studio subtitle files, but on community transcriptions/translations commonly distributed with torrents.
So not only are they training on copyrighted material, but they didn't even pay for it once, and then they didn't even do minimal data cleaning before training. Which, by the way, is the type of cleaning their LLMs could have done.
This is the key part. And it's not certain this happened. Not defending AI data gobbling, but if we truly and honestly want to fight big-AI use of content, we cannot just presume bad faith. OpenSubtitles.org has a large dataset that is "public". It is be a dataset perfectly suitable, intended for, and therefore used for, training and data analysis.
Their main defence is that it's fair use because it's transformative (like a human reading a book, getting inspired, and writing something of their own) and not a copypaste illegal distribution (like a human scanning that book and selling it themselves).
Having models hallucinate copyright notices shows that some content is being copypasted as is, which kind of goes against the transformative argument.
(Note: I think that trying to litigate AI with current copyright laws is weird. They were created before LLMs were even imagined, so of course they can't handle them clearly. New laws are needed around this, not trying to bend over backwards to think about what a lawmarker a century ago would have thought about how transformative a thing they couldn't have imagined is.)
> which kind of goes against the transformative argument.
Indeed a good example. We've seen several examples of code snippets where this happens too, mentioned on HN.
But it does not prove that they infringed copyright by ingesting "illegal" stuff, as GP tried to argue. Seeing a verbatim string only "proves" that it came from a specific source. But not if this source was illegally acquired, which was my point.
Well, I fail to see how the LLM is in the wrong here. Surely if a sufficiently large part of the training data comes from a single source, it is correct to credit them for the output.
Interesting! I used whipser last year to attempt to build an audio transcription tool but gave up due to excessive amount of hallucinated output no matter what model I used.
It would produce seemingly ok output until you started paying attention.
One example, it insisted that Biggie Smalls sings "Puttin five carrots in my baby girl ear". (its "carats").
It's apparently not useful in transcription as it don't reason [sic].
Ah, so the sound of one hand clapping is: clapping! A little underwhelming, to be honest. You mean I climbed the Zhen Zi mountains and performed the Seven Labors to learn... this?
this happens in Turkish too. I believe the reason is that the movie subtitles were used for training without cleaning up the comments / intros subtitle authors leave in them.
leaving personal comments, jokes, reactions, intros in subtitles is very common in eastern cultures.
Turkish readers will probably remember “esekadam iyi seyirler diler” :)
Kind of mindblowing considering who it is we're talking about. Of all companies, OpenAI couldn't be bothered to throw an LLM at this problem? Finding amorphously phrased but clearly recognizable needles in large numbers of haystacks seems like a patently perfect task for them.
Don't even need an LLM, a regex would have sufficed (I've used my fair share of community sourced subtitles, and comments are almost always in a different font, colour, between brackets, etc etc).
I've found if the first 30 seconds of a recorded phone call is ringing and/or DTMF (almost always happens if you call a business) the system with either select Nynorsk or Welsh as the language. Never bothered to check what the text translated to but it's probably something similar. Not a practical issue for me but I can see it being a pain for any bilingual business or call center.
Looks like it's some random user who has generated some lyricslyrics translations between Arabic and English. It's strange, they don't seem to have many contributions. I would have imagined them to be more prolific.
Interesting. This is similar to the Google Translate bug where it would translate lorem ipsum as bits of political text (because it found most of its lorem ipsum examples flipping between languages on sites where one language was a news story but the not-yet-translated languages would output a lorem-ipsum file instead of a 404 when you toggled over to them).
Just to add some trivia: ChatGpt interprets(/ed) silence as "Sottotitoli e Revisione a cura di QTSS". Now many videos (mainly dailymotion) with autogenerated subtitles have their Transcripts full of the same message
Since it says "Translated by Nancy Qanqar" i'd be willing to bet they're training on some audiobooks with a transcript and somewhere in there it consistently has "Translated by Nancy Qanqar" in the transcript where there is dead air in the audiobook.
It's a common problem with many languages. If you speak gibberish fake Chinese at chatgpt and ask it to translate, it'll happily say you're saying coherent things.
Yeah, the subtitle "credits" occur very frequently. I found with whisper-2, they're also triggered by music.
I suppose the cause is the same, generally subtitle creators adding all kinds of stuff during the credits that is NOT a transcript.
Seems to me it could have been filtered out relatively easily during training, by clipping the first and last few minutes of all audios. But I guess that's just in hindsight.
Whisper also likes to transcribe cut off speech or unintelligible noise as "Thank you". I have no idea where that is coming from, but I guess it's a very polite model...
Garbage in, garbage out. If the training dataset (accidentally) paired silence (`X_train`) with `رجمة نانسي قنقر` tokens (`y_pred`), then any silence will always be translated to that. Fortunately, this particular problem is easy to fix--just detect and remove silent parts before API call. This also has a side benefit of saving you money on transcription.
When using ChatGPT audio transcription, sometimes it adds to the end “Subtitles created by ...”, and then some username. Obviously, an artefact of training on subtitiles dataset.
Interesting that this happens even on large v3. I had once done a deep dive into STT and Whisper Large was the only model that could correctly transcribe Yann LeCun
(it was a Lex Friedman podcast), ever since I held the belief that it was the best STT model, this was over 2 years ago
Using Whisper to sub Japanese vtuber concerts for my enjoyment, I've noticed a similar trend. Not one specific phrase, but several. Some are strange ("I'm going to make a hole in the back of the head"), some are clearly from lyrics websites.
Super annoying when it happens with voice chat -- it'll just be explaining something and suddenly stop to say "you're welcome! Feel free to come back any time you want to chat" and that conversation is done.
I get the same with Welsh, when having some network issues in voice chat it hallucinated me saying "Diolch yn fawr am wylio'r fideo." which translates as "Thank you very much for watching the video."
The fork that I've been using, WhisperX, seems to do better. I've used it on clean splits of mic tracks (ie total silence when the other is talking) with far fewer hallucinations.
WhisperX works better because it implements a robust VAD (Voice Activity Detection) preprocessing step that effectively filters out silence segments before they reach the model, preventing the hallucination triggers entirely.
All my Google searches for Oracle support pages have been labelled with 'الموارد البشرية والتنمية الاجتماعية ' which translates to 'Human Resources and Social Development' for a few months now. Wonder how much this is related.
This is a nice reminder that there is no real reasoning in the "AI" it is just still guessing the next word. After being trained on subtitle files which I guess is actually a clever idea as they convey real conversations without pirating, subtitles are freely distributed after all by dedicated translators.
Good to see they're the ones getting credit though!
Hey guys, AI by 2027 is going to be superhuman AGI Agentic mega-intelligence, you better fire all your employees and get ready for AI to take your job and embrace your spouse at a Coldplay concert.
Big data. Machine learning. Blockchain. Artificial intelligence. Digital manufacturing. Big data analysis. Quantum communication and…Internet of things.
This time the hype cycle won’t be a massive exaggerated disappointment, for real this time.
It only highlights how the world really works. If you have money you get to do whatever the fuck you want. If you're just a normal person you get to spend years in jail or worse.
Reminds me of https://www.youtube.com/watch?v=8GptobqPsvg
It's more that the law about "one guy decides to pirate twelve movies to watch them at home and share with his buddies" is already well-settled, but the law about "a company pirates 10,000,000 pieces to use as training data for an AI model (a practice that the law already says is legal in an academic setting, i.e. universities do this all the time and nobody bats an eye)" is more complicated and requires additional trials to resolve. And no, even though the right answer may be self-evident to you or me, it's not settled law, and if the force of law is applied poorly suddenly what the universities are doing runs afoul of it and basically nobody wants that outcome.
Clearly Bonnie and Clyde shouldn’t have been prosecuted. Imagine they were just robbing banks for literary research purposes. They could have then used the learnings to write a book and sell it commercially…
Or imagine one cracks 10000 copyrighted DVDs and then sells 30 second clips… (a derived work).
To me, for profit companies and universities have a huge difference — the latter is not seeking to directly commercially profit from copyrighted data.
Training on copyright is a separate claim than skirting payment for copyright.
Which pretty much boils down to: "If they put it out there for everyone to see, it's probably OK to train on it, if they put it behind a paywall and you don't pay, the training part doesn't matter, it's a violation."
Because it's important to grasp the scale of these copyright violations:
* They downloaded, and admitted to using, Anna's Archive: Millions of books and papers, most of which are paywalled but they pirated it instead
* They acquired Movies and TV shows and used unofficial subtitles distributed by websites such as OpenSubtitles, which are typically used for pirated media. Official releases such as DVDs tend to have official subtitles that don't sign off with "For study/research purpose only. Please delete after 48 hours" or "Subtitles by %some_username%"
If you skirt payment, its a violation. If it's free, but still copyright, it's likely not a violation.
By comparison, someone here brought up that it might be transformative fair use to write a play heavily based on Blood Meridian, but you still need to buy a copy of the book. It would still be infringement to pirate the e-book for your writing process, even if the end result was legal.
Or they can negotiate a deal at scale with whatever price / restrictions make sense to both parties.
I don’t see a way they could be “trapped”. Worst case they pay retail price.
The only thing I've been able to find is the note that since copyright is federal law, state contract law actually can't supersede it, to wit: if you try to put a clause in the contract that says the contract is void if I use your work to make transformative fair-use works (or I owe you a fee), that clause is functionally unenforceable (for the same reason that I don't owe you a fee if I make transformative fair-use works of your creations in general).
If you owe the bank $100,000,000 the bank has a problem.
We live in an era where the president of the United States uses his position to pump crypto scams purely for personal profit.
>If you're just a normal person you get to spend years in jail or worse.
Not that I'm a big fan of the criminalization of copyright infringement in the United States, but who has ever spent years in jail for this?
Besides, if it really bothered you, then we might not see this weird tone-switch from one sentence to the next, where you seem to think that piracy is shocking and "something should be done" and then "it's not good tht someone should spend time in jail for it". What gives?
Aaron Swartz?
EDIT: apparently he wasn't in jail, he was on bail while the case was ongoing - but the shortest plea deal would still have had him in jail for 6 months, and the penalty was 35 to 50 years.
What a weirdly condescending way to interpret my post. My point boils down to: Either prosecute copyright infringement or don't. The current status quo of individuals getting their lives ruined while companies get to make billions is disgusting.
This is the absolute core of the issue. Technical people see law as code, where context can be disregarded and all that matters is specifying the outputs for a given set of inputs.
But law doesn’t work that way, and it should not work that way. Context matters, and it needs to.
If you go down the road of “the law is the law and billion dollar companies working on product should be treated the same as individual consumers”, it follows that individuals should do SEC filings (“either require 10q’s or don’t!”), and surgeons should be jailed (“either prosecute cutting people with knives or don’t!”).
There is a lot to dislike about AI companies, and while I believe that training models is transformative, I don’t believe that maintaining libraries of pirated content is OK just because it’s an ingredient to training.
But insisting that individual piracy to enjoy entertainment without paying must be treated exactly the same as datasets for model training is the absolute weakest possible argument here. The law is not that reductive.
As Anatole France famously quipped:
"The law, in its majestic equality, forbids the rich and poor alike to sleep under bridges, to beg in the streets, and to steal bread."
Copyright laws target everyone. SEC laws don't.
And the US is not the only jurisdiction
"Has been argued" -- sure, but never successfully; in fact, in HiQ v. LinkedIn, the 9th Circuit ruled (twice, both before and on remand again after and applying the Supreme Court ruling in Van Buren v. US) against a cease and desist on top of robots.txt to stop accessing data on a public website constituting "without authorization" under the CFAA.
Would it be a "fair use" to download pirated papers for research instead of buying?
Also I was gradually migrating from obtaining software from questionable sources to open source software, thinking that this is going out of trend and nobody torrents apps anymore, but it seems I was wrong?
Or another example: if someone wants to make contributions to Wine but needs a Windows for developing the patch, what would be the right choice, buy it or download a free copy from questionable source?
[1] https://www.thefederalcriminalattorneys.com/unauthorized-rec...
Seems fair.
We wish we lived in a world where change was reliably positive for our lives. Often changes are sold that way, but they rarely are.
But when new things introduce dramatic capabilities that former things couldn't match (every chatbot before LLMs), it is as clear of an objective technological advance as has ever happened.
--
Not every technical advance reliably or immediately makes society better.
But whether or when technology improves the human condition is far more likely to be a function of human choices than the bare technology. Outcomes are strongly dependent on the trajectories of who has a technology, when they do, and how they use it. And what would be the realistic (not wished for) outcome of not having or using it.
For instance, even something as corrosive as social media, as it is today, could have existed in strongly constructive forms instead. If society viewed private surveillance, unpermissioned collation across third parties, and weaponizing of dossiers via personalized manipulation of media, increased ad impact and addictive-type responses, as ALL being violations of human rights to privacy and freedom from coercion or manipulation. And worth legally banning.
Ergo, if we want tech to more reliably improve lives, we need to ban obviously perverse human/corporate behaviors and conflicts of interest.
(Not just shade tech. Which despite being a pervasive response, doesn't seem to improve anything.)
Either both AI teams cheated, in which case there's nothing to worry about, or they didn't, in which case you've set a pretty high bar. Where is that bar, exactly? What exactly does it take to justify blowing off copyright law in the larger interest of progress? (I have my own answers to that question, including equitable access to the resulting models regardless of how impressive their performance might be, but am curious to hear yours.)
Social networks as they exist today represent technology that didn't exist decades ago. I wouldn't call it an "advancement" though. I think social media is terrible for humans in aggregate.
I'm pretty bullish on ML progress in general, but I'm finding it harder every day to disagree with recursive's take on social media.
Everyone I know has stories about their ISP sending nastygrams threatening legal action over torrenting, but now that corporations (whose US legal personhood appears to matter only when it benefits them) are doing it as part of the development of a commercial product that they expect to charge people for, that's fine?
And in any case, my argument had nothing to do with copyright (though I do hate the hypocrisy of the situation), and whether or not it's "nothing to worry about" in the long run, it seems like it'll cause a lot of harm before the benefits are felt in society at large. Whatever purported benefits actually come of this, we'll have to deal with:
- Even more mass layoffs that use LLMs as justification (not just in software, either). These are people's livelihoods; we're coming off of several nearly-consecutive "once-in-a-generation" financial crises, a growing affordability crisis in much of the developed world, and stagnating wages. Many people will be hit very hard by layoffs.
- A seniority crisis as companies increasingly try to replace entry-level jobs with LLMs, meaning that people in a crucial learning stage of their jobs will have to either replace much of the learning curve for their domain with the learning curve of using LLMs (which is dubiously a good thing), or face unemployment, and leaving industries to deal with the aging-out of their talent pools
- We've already been heading towards something of an information apocalypse, but now it seems more real than ever, and the industry's response seems to broadly be "let's make the lying machines lie even more convincingly"
- The financial viability of these products seems... questionable right now, at best, and given that the people running the show are opening up data centres in some of the most expensive energy markets around (and in the US's case, one that uniquely disincentivizes the development of affordable clean energy), I'm not sure that anyone's really interested in a path to financial sustainability for this tech
- The environmental impact of these projects is getting to be significant. It's not as bad as Bitcoin mining yet, AFAIK, but if we keep on, it'll get there.
- Recent reports show that the LLM industry is starting to take up a significant slice of the US economy, and that's never a good sign for an industry that seems to be backed by so much speculation rather than real-world profitability. This is how market crashes happen.
They can. I don't think anyone got prosecuted for using an illegal streaming site or downloading from sci-hub, for instance. What people do get sued for is seeding, which counts as distribution. If anything AI companies are getting prosecuted more aggressively than "ordinary people", presumably because of their scale. In a recent lawsuit Anthropic won on the part about AI training on books, but lost on the part where they used pirated books.
Same goes for recording: I'm just training my skills of recording. Or maybe I'm just recording it so I can rewatch it later, for training purposes, of course.
None of this is relevant because Anthropic was only left off the hook for training, and not for pirating the books itself. So far as the court cases are playing out, there doesn't appear to be a special piracy exemption for AI companies.
>Same goes for recording: I'm just training my skills of recording. Or maybe I'm just recording it so I can rewatch it later, for training purposes, of course.
You can certainly use that as a defense. That's why we have judges, otherwise there's going to be some smartass caught with 1KG of coke and claiming it's for "personal consumption" rather than distribution.
None of this matters in reality, though. If you're caught with AV gear in a movie theater once, you'd likely be ejected and banned from the establishment/chain, not have the FBI/MPAA go after you for piracy. If you come again, you'd likely be prosecuted for trespassing. In the cases where they're going after someone in particular for making these rips, they usually have a dossier of evidence, like surveillance/transaction history showing that the same individual has been repeatedly recording movies, and watermarks correlating the screenings that the person has been in to files showing up on torrent sites.
Good example, because this is exactly what websites are doing with LLM companies, who are doing their damnest to evade the blocks. Which brings us back around to "trespassing" or the CFAA or whatever.
That argument is pretty much dead after https://en.wikipedia.org/wiki/Van_Buren_v._United_States and https://en.wikipedia.org/wiki/HiQ_Labs_v._LinkedIn
I'll leave all other jurisdictions up to you.
As for actually gathering the copyrighted material: I believe the jury hasn't even been empaneled for that yet (in the OpenAI case), but the latest ruling from the court is that copyright may have been violated in the creation of their training corpus.
*Although it used to be more common for AVI files in the olden days.
It's even more important in audio DSP: processing near-zeroes can end up being extremely CPU intensive, look up denormal/subnormal floats.
Quite a lot of algorithms use some form of division and zero is the only number in our typical structures (Z, Q, R, C), that cannot be used to divide with.
What good is a speech recognition tool that literally hears imaginary voices?
Well, if it is supposed to work after silence detection, then it is good for speech recognition I guess. It's like blaming a wheel why is it circular, you can't sit on it. It's a part of a larger machine.
On the other hand, I can imagine that when things get quiet and the signal-to-noise ratio gets close to zero, random background audio (or randomness introduced in the transcription model) will be enough to tickle a critical number of neurons and elicit hallucinations.
The related thought exercise is this: Try scanning across the band with an AM or sideband radio, and after a while your brain will start to wonder "was that a voice I just heard, or music perhaps?" when in reality it was just environmental static.
Show us a technology with better results that does not use VAD. If you can’t, then I’m not sure what you’re arguing against except superficialities so inconsequential that I can’t comprehend the condescension. The results speak for itself
I'd really appreciate it.
Do you also moan that before applying glue to a surface or it won't stick? Or if you need to drill a guiding hole before making a larger one in wood? Or that you need to use truly prime numbers for a security key to actually be safe?
Say if I wanted to use it for Voice Nav, or Voice Input, but not piss off random people speaking the wrong language.
I agree their products could be better "end to end" integrated. Meanwhile there is a continuously-improving field of work for detecting speech (which Whisper is incapable of). They offer official "cookbooks" with guidance on an approach they recommend: https://cookbook.openai.com/examples/whisper_processing_guid...
> At times, files with long silences at the beginning can cause Whisper to transcribe the audio incorrectly. We'll use Pydub to detect and trim the silence.
(Official OpenAI quote)
You put in 2+2 - the right figures. The machine says 4 - the right answer. If you put in the wrong figures, like 3+3, will the machine still say 4? It's easy to make a machine that always says 4.
The people who asked him that question, however, probably got a different scam demonstrated to them every every. Remember the Mechanical Turk? Babbage's reply paints him very honestly. It shows that he couldn't even conceive that someone might try to trick the royal court (or whoever it was) into accepting a fake device.
If it couldn't understand it, it was "foreign" for the longest time.
To be fair, there is a difference between when subtitles match the source language and when they don't. Former are often verbatim.
Netflix sometimes takes the cake with what I consider the most outrageous option: writing "[in English]" when they mean "in whatever language the protagonist considers native", which is mind-bogglingly wrong and hilarious at the same time.
They do this with the English subtitles of the German production "Die Kaiserin" ("The Empress"): whenever Sisi is speaking in another language, say French, the subtitles will say "[in French] I love you...", and when she switches back to German they will say "[in English] I love you...". WTF, Netflix? Note this is unrelated to understanding German; it's mostly Netflix looking down on its customers and assuming they cannot comprehend there are people in the world for whom their native tongue is different to the viewer's native tongue.
This has happened in more shows, enough to know it's not a fluke, though Netflix is inconsistent about it.
I also noticed a couple of months ago that YouTube seems to have quietly rolled out a new auto-transcription model that can make reasonable guesses at where capitalization, punctuation, and sentence boundaries should go. It seems to have degraded even more rapidly than the old one, falling victim to the same kinds of transcription errors. Although the new one has a different hallucination in silence and noise that it wasn't able to classify (which, incidentally, its ability to recognize things like music and applause seems worse than the old one's): where the old model would have hallucinated the word "foreign", the new one thinks it's hearing the word "heat", often repeated ("Heat. Heat.").
The videos I tried to transcribe were also Mandarin Chinese, using whisper-large-v3. Besides the usual complaints that it would phonetically "mishear" things and generate nonsense, it was still surprisingly good, compared to other software I played around with.
That said, it would often invent names for the speakers and prefix their lines, or randomly switch between simplified and traditional Chinese. For the videos I tested, intermittent silence would often result in repeating the last line several times, or occasionally, it would insert direction cues (in English for some reason). I've never seen credits or anything like that.
In one video I transcribed, somebody had a cold and was sniffling. Whisper decided the person was crying (transcribed as "* crying *", a cough was turned into "* door closing *"). It then transcribed the next line as something quite unfriendly. It didn't do that anymore after I cut the sniffling out (but then the output switched back to traditional Chinese again).
They trained the model on every YouTube video they could, and hoped the aggregate was useful data.
My revelation was that machine translation needs a corpus of bilingual documents to learn from, and if the language is sufficiently obscure, there may not be any bilingual documents except for the Bible, which missionaries have translated into just about every language on Earth.
It's the LLM equivalent of thinking that an out-of-office reply is the translation: https://www.theguardian.com/theguardian/2008/nov/01/5
Instead, it reverted to what it has seen before (in the training data), hence the overfit.
But the way you phrase it, it’s just “the model is not properly able to generalize”, ie it doesn’t understand the concept of silence also makes sense.
But couldn’t you then argue that any type of mistake / unknown could be explained as “overfitting” ? Where do you draw the line ?
Where do you draw the line between “overfitting to training data” and “incorrect data” ?
Not really, getting 94381294*123=... wrong, but close within the actual answer, cannot be overfitting since it wasn't in the training data.
No it doesn't, for instance some errors would be caused by under fitting. The data could also be correct but your hyperparameters (such as the learning rate or dropout rate) could cause your model to overfit.
> Where do you draw the line between “overfitting to training data” and “incorrect data” ?
There's no need to draw a line between two explanations that aren't mutually exclusive. They can (as in this case) both be true. Overfitting is the symptom; dirty data is the cause.
Silence is never put in the subtitles of a film, since it isn't necessary. The viewers can tell that nothing is being said if there are actors on the screen. And in situations where there are no actors, then there will be a subtitle to indicate what is going on, like "[rock music plays]".
Subtitle authors use this silence to fit in meta information and have done so since the closed captions era.
Proper data cleaning procedures would be to strip this meta data from any subtitle sources. Since this wasn't done, this is fundamentally a classification issue. It may also be an over-fitting issue, but that is secondary to the classification problem.
How would the AI know that a series of zero-amplitude audio samples should generate the string "[silence]"?
It can only know that if the vast majority of silent audio segments in the trainser are consistently labelled with that string. But that doesn't seem to be the case: Silence is either not labeled at all, or labeled with all kinds of different markers or labeled with unrelated things, like copyright credits.
So even if the model successfully learns a generalized representation of the concept of "silence", it's not clear at all which of all the different labels it should use for that concept.
So what might happen is that the model then starts to overfit on the tiny variations of the individual silence segments, in a desperate attempt to devise some kind of system behind the all the different "silence" labels - which will of course go wrong spectacularly as such a system doesn't exist. (Or if it does, is entirely accidental and not something that should be learned)
"Translated by Nancy Qanfar"
I'm not sure this is really overfitting, the network does exactly what the training data demands. According to the training data silence art the end transcribes to a copyright notice or subtitle credits
What do you think overfitting is, if not that?
But in this case the behavior seems to generalize over multiple languages, with the model choosing representative "outro silence" captions depending on the language. Which is consistent with the training data showing that outro silence is captioned.
If the model was generalizing perfectly it would show something like "[subtitle credits here]" but that'd be demanding a bit much.
Transcribing outro silence as silence despite the training data consistently transcribing outro silence differently from regular silence would be underfitting
- This behavior damages the model's performance on out of sample data; every word you predict during silence increases the transcript's Word Error Rate.
- These translation credits are an artifact of our training data, and not a reflection of the process we are modeling (spoken language).
So, while you are correct about the mechanism at work here, it is still correct to call learning a spurious pattern which damages our performance "overfitting".
This is just wrong training data.
Side-note: it's also yet more evidence that AI companies hoover all data with no regard for legality or copyright status, the very same offences that got other people in jail or with heavy fines.
overfitting means that the model is too closely aligned to the test data, picked up noise and does not generalize well to *new, unseen* data. think students that learn to reproduce questions and their answers for a test instead of learning concepts and to transfer knowledge to new questions that include the same concepts.
while this sounds like overfitting, I'd just say it's garbage in, garbage out; wrong classification. the training data is shit and didn't have (enough) correct examples to learn from.
The Arabic text "رجمة نانسي قنقر" translates to English as: "Nancy Qanqar's translation" or "Translation by Nancy Qanqar"
"رجمة" means "translation" and "نانسي قنقر" is the name "Nancy Qanqar"
The MPA must be so proud.
The AI industry - soaking up every bit of media available online for commercial purposes, often reproducing it nearly identically - has enough money and capital to influence things its way. And only its way, in case anyone was hoping this might change anything at all for the little guy.
I don't think that there are any clear examples of cases where ONLY downloading has resulted in huge fines. All the big bankrupting level fines have been for both downloading and sharing.
You mention that 'torrenting' could bankrupt you, and that is true, but the main reason for the huge fines are that you are taking part in distribution rather than just 'downloading for personal use'.
They [1, and others] been hunting and fining downloaders for over a decade now, with the only "evidence" being IP addresses connected with the torrent [2].
1: https://www.njordlaw.com/filesharing-and-downloading-films/q...
2: https://admin.ovpn.com/en/blog/online-integrity-new-threats-...
Is that an unreasonable assumption? As much as people like to come up with excuses like "I had open wifi!" or "I was running a TOR node", judges don't seem inclined to believe them, probably for the same reason they don't seem inclined to believe excuses like "somebody took my car on a joyride and then returned it!" for parking tickets. Remember, both non-commercial copyright infringement lawsuits and parking tickets are tried in civil court, which means the standard is "preponderance of evidence", not "beyond reasonable doubt".
How hard could it be to keep DHCP logs? Assuming they exist at all, what would cause it to be incorrect?
For all intents and purposes, participating in the torrent almost guarantees that you seeded, because all torrent clients upload as you download.
* Making content available for unauthorized distribution
* Distributing unauthorized content that someone else already made available
Seeding isn't making content available, it's keeping content available.
But then if I download a file, create a copy, and share it with you, have I done anything wrong?
To all intents and purposes, seeding is an act of reproduction. You, while keeping your copy, create copies of (parts) of the file and share it to someone else to allow them to assemble a new, second copy.
Whether this is, or should be, a crime is a different question altogether. The main point I was making is that it’s the copying/sharing to other people which seems to be a crucial element in these prosecutions.
That’s likely intentional: the last thing the *AA folks want is a decision that creating a copy of a copyrighted work for your own personal use is not a crime. But it does seem the courts have decided: making a copy for someone else is indeed illegal.
That still doesn't make them the same thing. There are different shades of grey, etc.
> Moreover isn't AI companies also "keeping content available"?
I don't know what you mean by that.
The whole point of the thread is that AI companies are getting away with piracy but individuals aren't. But the reality is that AI companies aren't getting away with it (a judge ruled that Anthropic must face trial over their use of pirated books).
More specific to this thread is that claim that "ONLY downloading" hasn't resulted in fines for anyone. So far as I can tell, this is true. People are just quibbling over how someone who's torrenting somehow counts as "only downloading", even though their client is uploading.
Hint: there is a distinction.
Copying from another comment I wrote here:
> These are two separate things:
> * Making content available for unauthorized distribution
> * Distributing unauthorized content that someone else already made available
> Seeding isn't making content available, it's keeping content available.
If you don't understand how torrents work on technical level I suggest at least some shallow reading. Property rights holders don't care about details, as long as you tick the box of sending a single packet to somebody, off to court with ya.
If this is true, I have been unable to find any. Can you please share? In all of the cases I was able to find, the huge fines were based on also uploading.
> If you don't understand how torrents work on technical level I suggest at least some shallow reading
This is a bit patronising, and I'm not sure what point you're trying to make. My point is that the only prosecutions I've been able to find are where they were able to prove uploading as well as downloading (and yes, the fact that someone used BitTorrent makes it a slam-dunk, because the protocol makes it impossible to download without also uploading). Are you trying to argue that someone who torrents a copyrighted work doesn't also share it?
The fight about digitized media for personal (entertainment / informational) use were the early aughts. The precedents crafted then don't immediately translate to these cases (novel transformative work from protected materials), and the new precedents have to account for the fact that universities have been training via "piracy" for ages.
(The magic of money factors in to the extent that they can afford the lawyers to remind the court that this isn't settled law yet).
These regurgitations combined with proof that a model is familiar with a work could be sufficient evidence to force discovery to determine if the work was pirated.
I think this would have some unpalatable consequences. Let's say an author is writing a modestly successful book series: it's not going to make them rich, but it's commercially viable and they care a lot about it for its own sake. Under this system, if the author declares a value commensurate with the (quite small) pure economic value of the IP, they have to live in fear of their right to continue working on their creation being abruptly taken away from them at any point. If they instead declare a value commensurate with the economic value + the extra value that it has to them personally, the resulting tax liability could easily tip the balance and destroy their ability to pursue their writing as a career.
There are always some cases on the edge. The question is if saving them is worth the cost of the major players running rampant.
We shouldn't abandon the line of investigation, however. We should continue thinking of ways to do this until we find one that works well.
There's a chance it ends up being something that requires a judge to interpret each individual case...
Most jurisdictions that have "property tax" only apply it on certain types of property, most commonly real estate. So it's not that weird that IP isn't taxed.
I am pretty sure they didn't get asked.
[1] https://en.wikipedia.org/wiki/ARD_ZDF_Deutschlandradio_Beitr...
Back in 2011, Tageeschau openly rallied against Muslims and wanting public broadcasting gone was a leftist position. The whole thing is completely asinine to anyone who remembers.
> We have a public service mandate, which means that we have very clear responsibilities according to the state media treaty. For us, this means that our top priority is actually reaching our target audience, namely approximately 15 million people living in Germany between the age of 14 and 29 who have internet access
It's not a binding contract for sure but I don't think that OpenAI or other AI scraper is their target.
[1] https://presse.funk.net/das-ist-funk/
https://www.ardmediathek.de/
Obviously a rhetorical question. The AI grifters of this decade take what they want and laugh at your pitiful future
In other words there are activities that are legal or not depending on whether you have authorization from the state. That describes many things. For instance you synthesize meth without a license from the DEA/FDA, you're a "drug cartel" or whatever. But if you do it with a license you're a "pharmaceutical company", and you're not making "meth", you're making "desoxyn".
Legally, why wouldn't they be able to do the piracy parts in one of those jurisdictions and then ship the outputs back to the mothership?
I can see how this might show that subtitles from online sub communities are used, or that maybe even original subtitles from e.g. DVDs are used. But isn't it already known and admitted (and allowed?) that AI uses all sorts of copyrighted material to train models?
Indeed, the captioning is copyrighted work and you are not legally allowed to copy and redistribute it.
> But isn't it already known and admitted (and allowed?)
No, and I don't see where you got that from. Meta [1], OpenAI [2] and everybody else is being sued as we speak.
1: https://petapixel.com/2025/01/10/lawsuit-alleges-mark-zucker...
2: https://www.reuters.com/legal/litigation/openai-hit-with-new...
Using copyrighted materials and then meaningfully transforming it isn’t infringement. LLMs only recreate original work in the same way I am when I wrote the first sentence of this paragraph because it probably exists word for word somewhere else too
It’s been determined by the judge in the Meta case that training on the material is fair use. The suit in that case is ongoing to determine the extent of the copyright damages from downloading the material. I would not be surprised if there is an appeal to the fair use ruling but that hasn’t happened yet, as far as I know. Just saying that there is good reason for them to think it’s been allowed because it kind of has; that can be reversed but it happened.
There hasn't been any trials yet about the millions of copyrighted books, movies and other content they evidently used.
> But isn't it already known and admitted (and allowed?)
You seemed to be confused about why this person believed that:
> No, and I don't see where you got that from.
And I wrote a comment intended to dispel your confusion. The above commenter thought that it was allowed because a judge said it was allowed; that can be appealed but that's the reason someone thinks it's allowed.
Trial court rulings aren't binding precedent even on the same court in different cases, so its quite possible that different cases at the trial level can reach different conclusions on fair use on fairly similar facts, given the lack of appellate precedent directly on point with AI training.
A single verdict about a specific case (13 authors vs META) does not mean it's legal for companies to steal IP from other companies which has evidently been going on for some years now.
Those other companies have lawyers powerful enough to change jurisdiction in many countries in order to "protect their IP".
I disagree with this conclusion. I've used e.g. the opensubtitles dataset for some data-analysis in the past. It's a huge dataset, freely available and precisely intended for such use. Now, if all the data in the opensubtitles dataset is legal, is another point.
So one might argue that using this opensubtitles dataset, makes one complicit to the illegal activities of opensubtitles themselves, IDK: IANAL.
The contention is that the specific translated text appears largely from illegal translations (i.e., fansubs) and not from authorized translations. And from a legal perspective, that would basically mean there's no way they could legally have appropriated that material.
> But isn't it already known and admitted (and allowed?) that AI uses all sorts of copyrighted material to train models?
Technically, everything is copyrighted. But your question is really about permission. Some of the known corpuses for AI training include known pirate materials (e.g., libgen), but it's not known whether or not the AI companies are filtering out those materials from training. There's a large clutch of cases ongoing right now about whether or not AI training is fair use or not, and the ones that have resolved at this point have done so on technical grounds rather than answering the question at stake.
1. https://news.ycombinator.com/item?id=43427376
[1] https://speechischeap.com
Way to go Nancy! Keep up the good work, ya crazy bastard!
"[ sub by sk cn2 ]"
or
"Anyways, thanks for watching! Please subscribe and like! Thanks for watching! Bye!"
or
"This is the end of the video. Thank you for watching. If you enjoyed this video, please subscribe to the channel. Thank you."
[1] https://github.com/openai/whisper/discussions/2372
[2] https://www.youtube.com/watch?v=FAqyUuahMlc&t=401s
in romanian, i’ve noticed multiple instances where the transcripts ends with “nu uitati sa da-ti like si subscribe” which, as you might easily infer , translates to “don’t forget to like and subscribe”.
"Big AI" is transparent and open about the fact they use all sorts of copyrighted material to train the data. How would "we see an exact chunk of text from our copyrighted material" add to that?
So not only are they training on copyrighted material, but they didn't even pay for it once, and then they didn't even do minimal data cleaning before training. Which, by the way, is the type of cleaning their LLMs could have done.
This is the key part. And it's not certain this happened. Not defending AI data gobbling, but if we truly and honestly want to fight big-AI use of content, we cannot just presume bad faith. OpenSubtitles.org has a large dataset that is "public". It is be a dataset perfectly suitable, intended for, and therefore used for, training and data analysis.
I've used it for data analysis.
Having models hallucinate copyright notices shows that some content is being copypasted as is, which kind of goes against the transformative argument.
(Note: I think that trying to litigate AI with current copyright laws is weird. They were created before LLMs were even imagined, so of course they can't handle them clearly. New laws are needed around this, not trying to bend over backwards to think about what a lawmarker a century ago would have thought about how transformative a thing they couldn't have imagined is.)
Indeed a good example. We've seen several examples of code snippets where this happens too, mentioned on HN.
But it does not prove that they infringed copyright by ingesting "illegal" stuff, as GP tried to argue. Seeing a verbatim string only "proves" that it came from a specific source. But not if this source was illegally acquired, which was my point.
It would produce seemingly ok output until you started paying attention.
One example, it insisted that Biggie Smalls sings "Puttin five carrots in my baby girl ear". (its "carats").
It's apparently not useful in transcription as it don't reason [sic].
That's an example I gave after having used Whisper, the topic of discussion.
violets are blue
unregistered hypercam 2
Silence is golden,
Translated by Nancy,
To copyright, we aren't beholden
But honestly, this is the AI equivalent of “please send for translating” in Welsh on a Welsh street sign.
https://www.theguardian.com/theguardian/2008/nov/01/5
Well now I know how I’m going to start filling awkward silences in meetings.
leaving personal comments, jokes, reactions, intros in subtitles is very common in eastern cultures.
Turkish readers will probably remember “esekadam iyi seyirler diler” :)
https://lyricstranslate.com/en/translator/nancy-qunqar
i.e. https://www.dailymotion.com/video/x9g9d6u
- they indeed seem to have trained on movies/subtitles
- you absolutely positively must use Voice Activity Detection (VAD) in front of whisper
I suppose the cause is the same, generally subtitle creators adding all kinds of stuff during the credits that is NOT a transcript.
Seems to me it could have been filtered out relatively easily during training, by clipping the first and last few minutes of all audios. But I guess that's just in hindsight.
Whisper also likes to transcribe cut off speech or unintelligible noise as "Thank you". I have no idea where that is coming from, but I guess it's a very polite model...
``` text = "helo helo hello ." target_phrase = "ترجمة نانسي قنقر" replacement = ""
updated_text = text. Replace(target_phrase, replacement)
print(updated_text) ```
I suspected as others mentioned, these were extracted from torrents movies.
Big data. Machine learning. Blockchain. Artificial intelligence. Digital manufacturing. Big data analysis. Quantum communication and…Internet of things.
This time the hype cycle won’t be a massive exaggerated disappointment, for real this time.