How AI Theft is Killing the Internet

an AI company stole my work and I'm far from the only one over recent months there's been a veritable Avalanche of Revelations about the unauthorized use of artistic journalistic and video works to train generative AI tools such as open AI chat GPT Google's Gemini and anthropics Claude where once there was merely a mountain of suspicion that such theft was taking place now sits a stack of cold hard evidence that some of the most valuable comp companies in the world right now have been developing their tools on the backs of the stolen labor of video creators journalists artists and authors A lot's been said about the legalities and morality of all of this but I want to talk about its broad a potential impact on the online media and Creator economy because even at a point where the AI stock bubble looks set to burst the rise of generative AI looks set to significantly alter who gets to PR profit from The Works of writers journalists artists and others and this has the potential to not only short change individual Publishers and creators but have a desperately damaging impact on who's able to contribute to our popular discourse and debate the future is AI to understand and create language it understands prompts written the way people write copyright Shield means that we will step in and defend our customers and pay the cost incurred is there I boom actually a bubble two major copyright lawsuits against AI art generators a couple of months ago my agent got in touch they were giving me advanced notice of an article which was about to be published by a recently launched news startup called proof news the piece in question was the result of an investigation carried out by journalists Annie Gilbertson and Alex Ryner into the use of YouTube videos to train the AI models which power generative AI tools including anthropics AI Chann bot Claude and Apple's new Apple intelligence system see as you probably know by now generative AI that is AI tools which responds to user inputed prompts by producing text images or video require huge SES of data to train generally speaking the more data or a tool can be trained on the better results it will produce and there's been a widely held assumption that the companies developing these tools have been freely helping themselves to the work of creators journalists artists and others in order to improve their products these AI companies have been pretty cryptic about what bodies of work they've been using for this purpose and the suspicion upon many has been that this is because they've been helping themselves to huge quantities of copyrighted material without asking for permission or giving any kind of payment to the creators of that work but Gilbertson and Ryner the writers of that article wanted proof which you know is kind of fitting given the name of the outlet they work for they thus began to dig through the research papers which developers often publish following new breakthroughs and when they did they found repeated references in Publications from anthropic apple and Nvidia to something called the pile the pile is a data set developed by a nonprofit organization called alther AI it weighs in at a whopping 825 gab of data which given it's comprised solely of text is absolutely wild comprising books Wikipedia articles Forum posts email chains and much much more the pile is an absolute Treasure Trove of written English it's perhaps no surprise then that anthropic apple and Nvidia have all made use of it in training their respective AI models as they trolled through the pile however Gilbertson and Ryner found something else they discovered that around 6 GB and 500 million words of this massive text vile was made up of subtitles scraped from YouTube videos the exact videos which had been stolen for the data set had been obscured but when they spotted that the data set included references to the video ID codes which are contained within YouTube links ryer was able to reverse engineer the data set to create a tool through which creators and their fans can find out whether their work had been used and when I searched my name I indeed found that that 18 of my own videos were among them now I was far from the worst affected Creator here some channels such as Ted's Ted Ed Channel had hundreds of videos stolen and the response to the proof news article from creators was unsurprisingly one of Despair Scott over at nerdsync made a quick response video which brimmed with Unapologetic rage Jacob Geller described the Revelation as nauseating and Abby from philosophy described a similar sense of violation I felt a little less personally wounded by all of this perhaps it was because the videos were old and thus the memory of the work that I put into them has mostly faded maybe it's because my videos tend to be less personal and vulnerable than those of other creators more than anything though I think I'm just used to this kind of stuff already I've been working in Creative Fields most of my life and particularly when it comes to making and distributing work over the internet I think I've just come to accept that someone somewhere is usually ripping me off in one way or another I didn't like it but I have come to accept that this kind of thing will happen even so this Revelation did set alarm bells ringing in the back of my head about the ways in which the Embrace of AI tools such as Claude or chat GPT looks at to reshape the economics of the creative portions of the web particularly when it comes to journalism commentary and other forms of social and political discussion one of the promises of the early internet was that it would give us a space in which we could more freely discuss the world around us many of us hoped it would allow us to come into contact with New Perspectives and voices and to some extent this has been the case even just on YouTube I'm constantly confronted with new outlooks and viewpoints which would never have been given airtime or page space in traditional media but what if this shift to an AI centered media economy is about to place that at risk what if the widespread theft of the work of journalists commentators and creators and its subsequent remixing and regurgitation by AI tools has the potential to place new limits on the kinds of perspectives we're able to come into contact with because there's a real danger that this could happen and not for the reasons that you might initially think woke AI machine super woke woke in the negative connotation of that word combat woke AI see there's been plenty of concern from right-wingers that AI models like chat GPT or mid Journey might impede political debate by being too woke the accusation is that developers might be so Keen to avoid any controversy or offense that they'll set these systems up to only give right on happy clappy liberal answers to users prompts which isn't an entirely unfounded concern Google recently had to pause the roll out of image generation on its Gemini chatbot after requests for it to produce pictures of German soldiers from World War II returned a plethora of racially diverse Nazis which is pretty historically inaccurate in a predictably ludicrous response Elon Musk decided to fix this dilemma by instructing developers of x's grock chatbot to program it to avoid answers that are woke this kind of institutional bias whether woke or in musk's case anti-woke is certainly a real challenge but there's a slightly more complex threat which emanates from how AI risks reshaping the economic model which currently funds the production of news related media because the Embrace of tools which have been set up to instantly scoop up any form of journalistic political or critical expression in order that they can be repurposed to generate profit for AI companies and their customers places the entire economic model on which such work relies at risk and where the internet has previously had at least something of a democratizing effect on the media landscape such a shift has the potential to return considerable power into the hands of corporate media magnetes now the internet has never been particularly kind to journalism the site now reports having 100 million videoos videos and claims 60,000 new videos are posted every single day one journalist typed relevant words into the internet one website alone the consumer Action Group has 113,000 members I mean I was actually amazed to learn 7 million people use the internet as the worldwide web started gaining widespread adoption during the 1990s more and more newspaper Publishers began launching websites for their titles the New York Times The Wall Street Journal and the Washington Post all began publishing daily articles on newly minted websites in 1996 while the guardian followed suit 3 years later doing so had immediate tradeoffs making their reporting freely accessible online surely meant that some people who otherwise would have paid to purchase a physical newspaper no longer would nevertheless the Hope among Media Company owners was that publishing online would massively increase their newspapers reach not only might some readers who previously would have been fiercely loyal to a rival publication give them a chance but they could reach international Readers and others for whom buying a physical copy of their paper might not even have been an option where going online meant for going upfront sales revenue then the general Hope was that additional advertising might make up the deficit a slide deck produced for starford the guardian at the time revealed their hope to become the intelligent person's home on the net the Assumption was that internet users who wanted to find out what was going on in the world on a given day would log onto the internet head to the homepage of of their paper of choice and browse around a bit reading a whole bunch of Articles and think pieces as they did they would be served with a veritable Feast of ads which would fund that journalism and line the pockets of each publication's owners that however is not the direction that things went in instead two new phenomena emerged which would fundamentally alter how people engage with the news on the one hand search engines and on the other social media from the very beginning Google stated that its goal was to organize the world's information and make it universally accessible and both Google and social media have played a key role in making the internet less for booding long gone of the days where one had to remember the URLs of the various websites you wanted to catch up on in fact to many people even typing in the URL of a website other than Google Facebook or Twitter now seems kind of archaic it's these services not the likes of the guardian or the Washington Post which would become our metaphorical homes on the net even for folks like me who boot up our phones or computers with the explicit intention of catching up on the news few navigate directly to newspaper websites most will instead load up social media to see what stories people are sharing And discussing there which as I mentioned before has had huge benefits for media plurality it makes it much easier and more natural to engage with mult multiple perspectives on a single news story in a way which would have been much less likely or at the very least much more expensive in the print era the challenge is that this has also made news publishing a much more precarious game a world in which readers simply dip into a news site to read a single article and then instantly return to social media is one in which Publications aren't able to show as many ads to each reader the associated reduction in reader loyalty also means there's less certainty in terms of how many people are going to visit each day and in the associated Revenue which comes from that what's really crucial to understand however is that this doesn't mean that news reporting journalism and other forms of current affairs commentary generate less economic value than it once did journalism continues to be incredibly lucrative the real shift is that a significant chunk of the value it creates now get sucked up by Google Facebook and other social media companies in pre- interet times newspapers could charge Hefty sums to companies wishing to advertise on their front pages but the role of the newspaper front page has effectively been replaced by the news feeds and results pages of thirdparty companies who now get to pocket that cash for themselves there's two ways to view this search and social media companies will argue that they deserve that money Google boasts that its products give people choice and help them find more highquality journalism than ever before they would argue that they are delivering new readers to news Publications and deserve to be financially rewarded for that the alternative perspective is that the key appeal of Google Facebook and the rest is massively boosted by the fact that users can turn to them as a portal to access the news one recent study suggested that Google searches for news stories generate the company $440 million US every year in this view Google is effectively using the work of journalists and their employers as free content for their sites newspapers thus find themselves in a position where they are generating significant value for big tech companies whilst they themselves struggle for cash some media organizations have sought to reassert their revenue generating abilities by introducing paywalls with some success where articles remain free to access several governments have begun to step in to protect Publishers in 21 the Australian government introduced the news media bargaining code this forced Google and Facebook to hand over around $50 million us to a variety of large media organizations in return for being allowed to list their content on news feeds and in search results nevertheless this is a mere Slither of the 11. 9 billion worth of Revenue which researchers at the center for economic policy research have suggested tech companies are effectively scalping from media companies each year the consequences for the quality of reporting that we rely upon to learn about the world around us are plain to see in a world where every single article has to fight to be read journalists are further encouraged to frame their reporting in sensationalism and clickbait where this hasn't helped to increase Revenue layoffs have been the order of the day 8,000 jobs were lost across the media Industries in the US UK and Canada in 2023 alone so bad of things got that the UK's Press Gazette now keeps a regularly updated page of redundancies but what does all of this have to do with AI well in effect what we've been left with is a system in which a huge chunk of the economic value created by journalists is scooped up by a new breed of intermediary companies who have inserted themselves between Publishers and readers I've discussed this model of what the Canadian writer Nick cernik calls platform capitalism in a handful of videos before and again there have been some positives to come out of this so far it's maybe worth acknowledging that you probably wouldn't be watching this video If YouTube hadn't emerged as a platform which connects people like me who create news related videos with potential viewers regardless of what you think of the quality of my work or that of any other creators this has allowed all kinds of perspectives and opinions which would never find a voice in Legacy Media to get a fair Shake nevertheless the rise of AI tools such as chat GPT and the content theft which makes those tools possible will only serve to turbocharge the power of these new intermediary companies and the way in which it will do so risks crushing this more positive aspect of the New Media economy it's certainly not all good news for media Barons such as Rupert Murdoch or Jeff Bezos either but there's a real danger that adds a further twist to this story in which the power of such figures to shape the news that we read watch and listen to is stronger and more Insidious than ever before large language models have captured the world's imagination this is the Humane AI pin about a year ago November 30th we shipped chat GPT Apple intelligence is the personal intelligence system that puts powerful generative models right at the core of your iPhone iPad and Mac ever since the first mainstream generative AI tools were announced in 2022 we've been encouraged to view the likes of chat gbt Claude Gemini and mid Journey as completely new and revolutionary and on a purely technical level it would be ludicrous to suggest that they're not pretty impressive that being said they didn't come from nowhere and they're really much more of an incremental step forward from previous machine learning Technologies than some kind of pure unpredictable invention the same is true of of the economic impact AI is likely to have on our online media ecosystem as we've seen the trajectory of the digital economy today has been one in which the economic value created by journalists belong with that created by visual artists video creators and others has increasingly been siphoned off by big tech companies these companies have inserted themselves as intermediaries between content providers on one hand and audiences on the other and expect to profit handsomely from allowing the former to share their work with the latter generative AI tools and the people and companies who use them to create new works are effectively another form of intermediary let's say that I decide to launch an AI powerered News website and because I want to attract a young heavily online audience I decided to publish as my first article a 1,000w piece about Charlie XCX declaring that carela Harris is brat now I'm not a young man anymore I've not listens to Charlie xcx's album brat I don't know what a brat summer is and given it's now Autumn it's probably too late for me to find out but that doesn't matter because I'm not actually writing the article chat GP tears although in reality chat GPT isn't writing the article either instead some of the article substance will be drawn from chat GPT searching the internet and some of it will be produced by calculations the all makes based B on its prior training the resulting article is actually a Frankenstein like amalgamation of perhaps hundreds of previously published articles some of those will explicitly be about KLA Harris Andor Charlie XCX whilst other articles and books and yes YouTube video subtitles will have been used in a more broader sense to teach chat GPT how to construct a decent sentence and paragraph and argument in short then chat GPT is acting as an intermediary it's taking a huge sample of pre-existing work written by journalists music reviewers social media users and others and simply rewriting it for a new audience what makes this far worse than the large chunks of Revenue which Google YouTube Tik Tok and Facebook take from Publishers is that neither open AI as the creator of chat GPT nor I as the publisher of this fictional website are sharing any of the revenue we make from subscriptions or advertising with those original authors at all in fact they're not even getting the basic decency of credit for their work whether or not this is legal it's obviously immoral chat GPT might olude this act of plagiarism by combining an unknowable number of sources and processing them through a mystery box of maths and code but it and I would still be taking the work of others and presenting it even not as my own than as the work of chat GPT whatever hurt or damage this individual example might cause however the broader consequences of such practices becoming widespread are even more Troublesome because the bigger question to ask is this in a world where anything that someone writes and posts on the internet including journalists can quickly be gobbled up by a generative AI tool regurgitated back out and republished elsewhere that's not only going to lead to a handful of individual writers or news Publishers being short changed it has the potential to make original writing or indeed video production or podcast development completely financially unsustainable journalism in particular can be really expensive to produce I can tell you firsthand from my experience making videos for this channel that the cost of researching and writing and editing and arranging interviews and traveling to any necessary locations to report on events adds up incredibly quickly I'm also pretty lucky in this regard my work and that of most other creators and Publishers who make journalism adjacent content for YouTube largely revolves around analyzing topics and events which other far more qualified journalists have already reported on producing breaking news is endlessly more expensive it requires reporters to have the time to go to press conferences and knock on doors and interview politicians and visit crime scenes and submit Freedom of Information requests and monitor email inboxes in case anyone gets in touch with a lead on a story which no one else even knows about yet it's not even particularly reliable in terms of producing a monetizable end product a reporter might spend ages working on a story only for it to not really go anywhere of course the discursive nature of Journalism and current affairs commentary means that no writer or publication is ever able to monopolize every single penny of the e iic value that each of their stories generates even when a reporter has the honor of breaking an entirely new story or when an opinion writer offers a particularly spicy new take on a topic that story or analysis is often quickly picked up by other journalists and Publications who want to discuss it too that's all an accepted part of the game but an AI fueled media economy in which any article video or audio piece can be instantly sucked up by an AI tool Reg itated into a new form and then republished elsewhere is going to place the very Financial viability of all forms of Journalism into question at best such a shift will likely mean that even the most well-meaning journalists and writers will have to cut Corners they'll have less time to dig beneath the surface of a given story to see What Lies Beneath this would only further embolden politicians and PR companies to fill their announcements and press releases with more questionable claims knowing that overworked and underpaid journalists won't have the resources to fully investigate whether or not they're true at worst it has the potential to kill off small independent news outlets entirely leaving behind only the handful of large Legacy Outlets which have dominated our political discourse for more than a century see over the past year news organizations have begun to fight back against the impact AI tools have been having on their revenues in December 2023 the New York Times began the process of suing open AI for allegedly using its articles to train the AI model behind chat GPT terrified that it might face further lawsuits open AI began to make deals with other Publishers including rert Murdoch's news corporation which owns The Wall Street Journal and the Sun and K Nast which owns the New Yorker Vogue and several other titles the former deal was reported to be worth in the region of 250 million which might seem like the beginnings of a more Equitable model for training AI tools whilst also better reimbursing journalists and Publishers for their work the reality however is that such deals are likely to remain The Preserve of large corporate media organizations with both illegal heft and political connections required to force AI companies into making such Arrangements there will have been plenty of Journalism adjacent creators and channels in the training data stolen from YouTube and used to train Claude ble intelligence and nvidia's AI models but regardless of whether those companies stealing those video transcripts was legal or not most of them simply won't have the ability to act on these revelations in order to secure a payout in short then the impact of the AI ification of our public discourse will not be shared equally established Publishers will be able to defend themselves and their revenue streams from The Fallout whilst without much broader legislative intervention independent creators Publishers and smaller Outlets will largely have to accept their work being routinely stolen of course some writers podcasters creators Etc might be Saved By Their audience having enough of a personal connection to them as personalities that they continue to engage with their work even if it could technically be found elsewhere but this in itself presents new challenges a popular discourse based on personalities and parasocial rather than ideas isn't necessarily a great one others might continue to publish for fun or out of a sense of civic duty nevertheless their ability to grow the scale of what they do and improve its quality is likely to be severely hindered the result is that the more negative aspect of the New Media revolution in which intermediary companies skim revenue from writers and Publishers will be turbocharged while at the same time the more positive outcomes relating to Media plurality are rolled back we risk once again finding ourselves in a position where our Collective social political and cultural conversation is dominated by just a handful of corporate Outlets owned by an Ever diminishing and merging number of companies and Proprietors we might read their articles directly on their we sites or through an AI chatbot or articles plagiarized through their use but it will be those corporate media Outlets who be the only ones with the means to do the original reporting which Powers these tools and therefore who will still be shaping the assumptions and perspectives which these tools spit out that such a scenario would be devastating for our ability to think and engage critically with the world around us barely needs saying a healthy journalistic ecosystem needs more voices not less but if AI companies continue to be allowed to steal the work of independent writers journalists and creators of all Stripes without permission or as importantly compensation the ability of anyone without considerable financial and institutional backing to contribute meaningfully to our Collective discourse is going to be severely constrained thankfully some companies are beginning to reject the generative AI Invasion just last month the team behind the graphics app procreate pledged never to add generative AI capabilities to their software or to use their software to steal users work which seems like a ridiculous statement to have to make but Adobe found themselves in hot water in February and their new terms of service seem to imply that they would be doing just that another company which Has Come Out Swinging against a ey theft is nebula the Creator own streaming service that I've been a part of for some years now the publication of the proof news article that I mentioned at the beginning of this video was a real wakeup call for many creators on the platform and so the team opted to make a stand by decisively declaring that none of the platform's original Productions will use generative AI tools and what's particularly exciting for me at least is that the next of those original Productions to be released as of the uploading of the this video is my featurelength nebula original documentary Boomers if you're a regular viewer of my channel you'll probably heard me talk about Boomers already perhaps you've seen the announcement video which went out a couple of weeks ago it's a featurelength documentary which Dives head first into the Generation War which sometimes seems to have taken over our politics Boomers has been about a year in the making at this point and has involved me traveling to various spots around the US the UK and over to France in order to interview experts ordinary folks and even my own mom my goal has been to try and shed some light on the lives and Legacy of the baby boomer generation I wanted to understand why it is that the Boomers often seem to have soaked up all the wealth and power in our society and why everything seems to be so much more difficult for the generations which have come after them housing University tuition pensions the climate crisis it covers a lot and by some weird twist of fate also ended up involving me driving a golf buggy around a fake Spanish Town in Florida you'll have to watch the film to understand that one the fact that Boomers is a nebula original production has allowed me to be so much more ambitious with this project than anything I've ever made before this is a giant leap up from making these little YouTube videos into making a proper fullscale documentary I've worked with a hugely talented team of producers and camera operators and editors and production staff to make the whole thing look and sound absolutely gorgeous and I cannot wait to share it with you all extremely soon Boomers will be available to watch exclusively on nebula and if you're not signed up already now is a fantastic time to get on board over the course of this year we've really upped our game with an incredible slate of original films and series which like Boomers aren't available to watch anywhere else just recently we released Abigail Thorne of philosophy tub's fantastic new short film Drcula's ex-girlfriend we also recently announced a featurelength documentary by Bobby Brockley called 17 Pages which covers the Baltimore Affair a massive scientific Scandal retold through interviews and some really cool original artwork this is just a taste of some of the fantas fastic stuff which has been coming out on the platform lately all of which is allowing creators like me to push ourselves beyond the limits of YouTube and to make full professional films and series that we're really passionate about if you want to sign up for nebula you can get a cheeky little discount on doing so by using my personal discount Link at go. nebula.

com Nicholas that will give you 40% off an annual plan allowing you to watch Boomers and all of that other exclusive video goodness for just $3 per month that's so much cheaper than any other streaming service out there it's basically a no-brainer or if you don't want to add another monthly subscription to your budget we've also recently been experimenting with lifetime subscriptions which have proven really really popular these naturally cost more up front but give you access to nebula and all the great new stuff that we're making for as long as both you and the platform exist sign up using my personal link really really helps to support my work I'd love for Boomers to be the first of many large scale documentary projects that I'm able to work on and if lots of people sign up early then it helps to signal to the team at nebula that folks are interested in seeing what I make and will help me make the case for them to throw more time resources and money my way for some of the cool projects that I've got planned in the future that link again is go. neb. V slom Nicholas I'll see you over on nebula and we'll look forward to sharing the trailer for Boomers with you very very soon thank you so so much for watching this video I hope it has given you some food for thought around AI theft and let me know your thoughts uh down in the comments all that's left for me to do right now is to thank my top tier supporters over on patreon those supporters are agent Maxwell Alexander blank Alan Gan Amit sing paraha Bill Mitchell Christopher Cowan dickon Spain Fiasco Linguini Gabriel KO Gary Glenn sugden jaraba Jimmy Dunn Lisa Yuan Neila bugard paulus edicus Richard Richard rapoon Sergio Suarez Sophia R strange weekend yilson zedy Reese and Zoe Alden if you want to join them in getting copies of the scripts to these videos uh my weekly Blog the Friday update some early sneak peeks at stuff that we're working on and a bunch of other stuff stuff then you can find out how to do so over at patreon.