Elon Musk did not create an AI trained on your fanfiction.

via https://ift.tt/zbnHfvL

cfiesler https://cfiesler.tumblr.com/post/702901922538061824/elon-musk-did-not-create-an-ai-trained-on-your :

Hi, AI ethicist + fanfiction expert here. (This is one of those times where I feel uniquely qualified to comment on something…)

I’m seeing this weird game of telephone about the Sudowrite AI that I think started out pretty accurate, but now has become “Elon Musk created an AI that is stealing your fanfiction” (which frankly gives him far too much credit). I can probably say more about this, but here are a few things that I want to clarify for folks, which can be boiled down to “Elon Musk has nothing to do with this” and “this is nothing new”:

Elon Musk is not involved in any way with Sudowrite, as far as I can tell. Sudowrite does, however, use GPT-3, the widely-used large language model created by OpenAI, which Elon Musk co-founded. He resigned in 2018, citing a conflict of interest due to Tesla’s AI development. It wasn’t until after he left that OpenAI went from being a non-profit to a capped for-profit. Elon Musk doesn’t have anything to do with OpenAI currently (and in fact just cut off their access to Twitter data), though I can’t find anything that confirms whether or not he might have shares in the company. I would also be shocked if Elon actually contributed anything but money to the development of GPT-3.

Based on Sudowrite’s description on their FAQ https://www.sudowrite.com/FAQ, they are not collecting any training data themselves - they’re just using GPT-3 paired with their own proprietary narrative model. And GPT-3 is trained on datasets like common crawl and webtext, which can simplistically be described as “scraping the whole internet.” Same as their DALL-E art generator. So it’s not surprising that AO3 would be in that dataset, along with everything else (e.g., Tumblr posts, blogs, news articles, all the words people write online) that doesn’t use technical means to prohibit scraping.

OpenAI does make money now, including from companies like Sudowrite paying for access to GPT-3. And Sudowrite itself is a paid service. So yes, someone is profiting from its use (though OpenAI is capped at no more than 100% return on investment) and I think that the conversations about art (whether visual or text) being used to train these models without consent of the artist are important conversations to be having.

I think it’s possible that what OpenAI is doing is legal (i.e., not copyright infringement) for some of the same reasons that fanfiction is legal (or perhaps more accurately, for reasons that many for-profit remixes are found to be fair use), but I think whether it’s ETHICAL is a completely different question, and I’ve seen a huge amount of disagreement on this.

But the last thing I will say is that this is nothing new. GPT-3 has been around for years and it’s not even the first OpenAI product to have used content scraped from the web. (Your picture was not posted)

How monopoly enshittified Amazon

capitalism

via https://ift.tt/CM1Wesb

mostlysignssomeportents https://mostlysignssomeportents.tumblr.com/post/702184812110315520/how-monopoly-enshittified-amazon :

In Bezos’s original plan, the company called “Amazon” was called “Relentless,” due to its ambition to be “Earth’s most customer-centric company.” Today, Amazon is an enshittified endless scroll of paid results, where winning depends on ad budgets, not quality.

Writing in Jeff Bezos’s newspaper The Washington Post, veteran tech reporter Geoffrey Fowler reports on the state of his boss’s “relentless” commitment to customer service. The state is grim.

https://www.washingtonpost.com/technology/interactive/2022/amazon-shopping-ads/

Search Amazon for “cat beds” and the entire first screen is ads. One of them is an ad for a dog carrier, which Amazon itself manufactures and sells, competing with the other sellers who bought that placement.

Scroll down one screen and you get some “organic” results — that is, results that represent Amazon’s best guess at the best products for your query. Scroll once more and yup, another entire screen of ads, these ones labeled “Highly rated.” One more scroll, and another screenful of ads, one for a dog product.

Keep scrolling, you’ll keep seeing ads, including ads you’ve already scrolled past. “On these first five screens, more than 50 percent of the space was dedicated to ads and Amazon touting its own products.” Amazon is a cesspit of ads: twice as many as Target, four times as many as Walmart.

How did we get here? We always knew that Amazon didn’t care about its suppliers, but being an Amazon customer has historically been a great deal — lots of selection, low prices, and a generous returns policy. How could “Earth’s most customer-centric” company become such a bad place to shop?

The answer is in Amazon’s $31b “ad” business. Amazon touts this widely, and analysts repeat it without any critical interrogation, proclaiming that Amazon is catching up with the Googbook ad-tech duopoly. But nearly all of that “ad” business isn’t ads at all — it’s payola.

https://pluralistic.net/2022/02/27/not-an-ad/#shakedowns

Amazon charges its sellers billions of dollars a year through a gladiatorial combat where they compete to outspend each other to see who’ll get to the top of the search results. May the most margin-immolating, deep-pocketed spender win!

Why would sellers be willing to light billions of dollars on fire to get to the top of the Amazon search results?

Prime.

Most of us have Amazon Prime. Seriously — 82% of American households! Prime users only shop on Amazon. Seriously. More than 90% of Prime members start their search on Amazon, and if they find what they’re looking for, they stop there, too.

If you are a seller, you have to be on Amazon, otherwise no one will find your stuff and that means they won’t buy it. This is called a monopsony, the obscure inverse of monopoly, where a buyer has power over sellers.

But monopoly and monopsony are closely related phenomena. Monopsonies use control over buyers — the fact that we all have Prime — to exert control over sellers. This lets them force unfavorable terms onto sellers, like deeper discounts. In theory, this is good for use consumers, because prices go down. In practice, though…

Back in June 2021, DC Attorney General Karl Racine filed an antitrust suit against Amazon, because the company had used its monopoly over customers to force such unfavorable terms on sellers that prices were being driven up everywhere, not just on Amazon:

https://pluralistic.net/2021/06/01/you-are-here/#prime-facie

Here’s how that works: one of the unfavorable terms Amazon forces on sellers is “most favored nation” status (MFN), which means that Amazon sellers have to offer their lowest price on Amazon — they can’t sell more cheaply anywhere else.

Then Amazon hits sellers with fees. Lots of fees:

• Fees to be listed on Prime (without which, your search result is buried at the bottom of an endless scroll):
• Fees for Amazon warehouse fulfillment (without which, your search result is buried at the bottom of an endless scroll)

And finally, there’s payola — the “ads” you have to buy to outcompete the other people who are buying ads to outcompete you.

All told, these fees add up to 45% of the price you pay Amazon — sometimes more. Companies just don’t have 45% margins, because they exist in competitive markets. If I’m selling a bottle of detergent at a 45% markup, my rival will sell it at 40%, and then I have to drop to 35%, and so on.

But everyone has to sell on Amazon, and Amazon takes their 45% cut, which means that all these sellers have to raise prices. And, thanks to MFN, the sellers then have to charge the same price at Walmart, Target, and your local mom-and-pop shop.

Amazon’s monopoly (control over buyers) gives it a monopsony (control over sellers), which lets it raise prices everywhere, at Amazon and at every other retailer, even as it drives the companies that supply it into bankruptcy.

Amazon is no longer a place where a scrappy independent seller can find an audience for its products. In order to navigate the minefield Amazon lays for its sellers (who have no choice but to sell there), these indie companies are forced to sell out to gators (aggregators), which are now multi-billion-dollar businesses in their own right:

https://pluralistic.net/2022/02/10/monopoly-begets-monopoly/#gator-ade

This brings me back to the enshittification of Amazon search, AKA late-stage (platform) capitalism. Amazon’s dominance means that many products are now solely available on the platform. With the collapse of both physical and online retail, Prime isn’t so much a choice as a necessity.

Amazon has produced a planned economy run as capriciously as a Soviet smelting plant, but Party Secretary Bezos doesn’t even pretend to be a servant of the people. From his lordly seat aboard his penis-rocket, Bezos decides which products live and which ones die.

Remember that one of those search-results for a cat-bed was a product for dogs? Remember that Amazon made that dog product? How did that end up there? Well, if you’re a seller trying to make a living from cat-beds, your ad-spending is limited by your profit margin. Guess how much it costs Amazon to advertise on Amazon? Amazon is playing with its own chips, and it can always outbid the other players at the table.

Those Amazon own-brand products? They didn’t come out of a vacuum. Amazon monitors its own sellers’ performance, and creams off the best of them, cloning them and then putting its knockoffs above of the original product in search results (Bezos lied to Congress about this, then admitted it was true):

https://nypost.com/2021/10/18/jeff-bezos-may-have-lied-to-congress-about-amazon-practices-reps/

If you’ve read Chokepoint Capitalism, Rebecca Giblin’s and my new book about market concentration in the entertainment industry, this story will be a familiar one. You’ll recall that Amazon actually boasts about this process, calling it “the flywheel”:

https://twitter.com/rgibli/status/1561761732108107777

Everything that Amazon is doing to platform sellers, other platforms are doing to creators. You know how Amazon knocks off its sellers’ best products and then replaces them with its clones? That’s exactly what Spotify does to the ambient artists in its most popular playlists, replacing them with work-for-hire soundalikes who aren’t entitled to royalties.

You can learn more about how Spotify rips off its performers in the Chokepoint Capitalism chapter on Spotify; we made the audiobook version of that chapter a Spotify exclusive (it’s the only part of the book you can get on Spotify):

https://pluralistic.net/2022/09/12/streaming-doesnt-pay/#stunt-publishing

Entertainment and tech companies all want to be the only game in town for their creative labor force, because that lets them turn the screws to those workers, moving value from labor to shareholders.

Amazon is also the poster-child for this dynamic. For example, its Audible audiobook monopoly means that audiobook creators must sell on Audible, even though the #AudibleGate scandal revealed that the company has stolen hundreds of millions of dollars from these creators. (Our chapter on Audiblegate is the only part of our audiobook on Audible!)

https://pluralistic.net/2022/09/07/audible-exclusive/#audiblegate

Then there’s its Twitch division, where the company just admitted that it had been secretly paying its A-listers 70% of the total take for their streams. The company declared this to be unfair when the plebs were having half their wages clawed back by Amazon, so they fixed it by cutting the A-listers’ pay.

https://pluralistic.net/2022/09/22/amazon-vs-amazon/#pray-i-dont-alter-it-further

Twitch blamed the cut on the high cost of bandwidth for streaming. If that sounds reasonable to you, remember: Twitch buys its bandwidth from Amazon. As Sam Biddle wrote, “Amazon is charging Amazon so much money to run the business via Amazon that it has no choice but to take more money from streamers.”

https://twitter.com/samfbiddle/status/1572667269284777984

As Bezos suns himself aboard his yacht-so-big-it-has-a-smaller-yacht, we ask him to referee a game where he also owns one of the teams. Over and over again, he proves that he is not up to the task. Either his “relentless” customer focus was a sham, or the benefits of cheating are too tempting to ignore.

Historically, we understood that businesses couldn’t be trusted to be on both sides of a transaction. The “structural separation” doctrine is one of the vital pieces of policy we’ve lost over 40 years of antitrust neglect. It says that important platforms can’t compete with their users.

https://locusmag.com/2022/03/cory-doctorow-vertically-challenged/

For example, banks couldn’t own businesses that competed with their commercial borrowers. If you own Joe’s Pizza and your competitor is Citibank Pizza and you both have a hard month and can’t make your payment, will you trust that Citi called in your loan but not Citibank Pizza’s because they had a more promising business?

Today, all kinds of businesses have been credibly accused of self-preferencing: Google and Apple via their App Stores, Spotify via its playlists, consoles via their game stores, etc. Legislators have decided that the best way to fix this isn’t structural separation, but rather, rules against self-preferencing.

Under these rules, companies will have to put “the best” results at the top of their listings. This is doomed. When Apple says it put its own ebook store ahead of Bookshop.org’s app because it sincerely believes Apple Books is “better,” how will we argue with this? Maybe Apple really does believe that. Maybe it doesn’t. Maybe it does, but only because of motivated reasoning (“It is difficult to get a product manager to understand something, when their bonus depends on them not understanding it”).

The irony here is that these companies’ own lawyers know that a sincere promise of fairness is no assurance that your counterparty will act honorably. If the judge in Apple v. Epic was a major shareholder in Epic, or the brother-in-law of Epic’s CEO, Apple’s lawyers would bring down the roof demanding a new judge — even if the judge promised really sincerely to be neutral.

https://marker.medium.com/moral-hazard-and-monopoly-42e30eb159a8

Ultimately, it doesn’t matter if Amazon’s enshittification is because Bezos was a cynic or because he sold out. Once Amazon could make more money by screwing its customers, that screw-job became a fait accompli. That’s why it’s so important that the FTC win its bid to block the Activision-Microsoft merger:

https://www.politico.com/news/2022/11/23/exclusive-feds-likely-to-challenge-microsofts-69-billion-activision-takeover-00070787

The best time to prevent monopoly formation was 40 years ago. The second best time is now.

Anti-monopoly measures are slow and ponderous tools, but when it comes to tech companies, we have faster, more nimble ones. If we want to make it easy to compete with Amazon, we could — for example — use Adversarial Interoperability to turn it into a dumb pipe:

https://pluralistic.net/2022/08/01/dumb-pipes/#original-asin

That is, we could let local merchants use Amazon’s ASIN system to tag their own inventory and produce a realtime database. Customers could browse Amazon to find the things they wanted, with a browser plugin that turned “Buy It Now” into “Buy It Now at Joe’s Hardware”:

https://doctorow.medium.com/view-a-sku-32721d623aee

But this only works to the extent that Amazon’s search isn’t totally enshittified. To that end, Fowler has a few modest proposals of his own, like requiring that at least 50% of the first six screens be given over to real results, not ads.

“Perhaps 50 percent sounds like a lot to you? But even that rule would force Amazon to show us at least some of the most-relevant results on the first screen of our device…Amazon wouldn’t comment on this suggestion.” (Your picture was not posted)

dreamwidth

via https://ift.tt/pRuC5nv

so I lost my login to my dreamwidth and haven’t managed to get it re-set, but I can still crosspost to Tumblr. It seems that not everyone reading my Dreamwidth feed is aware that the posts there are largely reblogs from Tumblr, and thus are not original compositions by me, and I do not mean to represent them as such– it is not possible for me to customize the IFTTT routine I use for crossposting in such a way as to make it more plain. But there is always a link back to the original Tumblr post, and I apologize if that is not totally obvious to anyone reading this on Dreamwidth. I have a whole fucking situation about my password managers that I just haven’t sorted out, and can’t, and this is the best I can do. Again, if something has come through that starts with a link back to a Tumblr post and has some weird and confusing formatting, if you click through it’s gonna be obvious that it’s a reblog, and I am not intending to represent it as something I have originally created. It’s just not possible to make it more obvious, with any of the technological solutions I am actually capable of using. (Your picture was not posted)