ArsTechnica claims privacy rights are importante but we can’t delete our accounts.

[SuperDave]

Where others see The Downfall Of All We Hold Dear, I'm seeing opportunity. It wouldn't take a really significant plurality of the Internet, if the sources were sufficiently widespread and sufficiently committed, to turn AI into steampunk orientation.

The one thing the Internet has produced more better than anything which came before is trolls. I'm kind of looking forward to seeing what the trolls do with this once they realize the opportunity. It'll be fun to watch. A sort of new DDoS....

 

[von Chaps]

Where others see The Downfall Of All We Hold Dear, I'm seeing opportunity. It wouldn't take a really significant plurality of the Internet, if the sources were sufficiently widespread and sufficiently committed, to turn AI into steampunk orientation.

That's all fine and good, but to poison the well, you need to actually... poison the well - i.e. piss in your own chips. What value would the forums have when they are full of trollish garbage?

In fact, the value proposition of the forums is going to decline anyway as people become more circumspect about what they post. I am certainly less inclined now to try and add any value to my posts by including factual references and research. I don't pay to post here so that I can work as a researcher for Condé Nast or OpenAI gratis.

So actually, you are probably right. The signal to noise ratio here is going to go down. And that's a problem.

I'm not rage-quitting (that's not in my nature), but I do have a long memory and I will probably just wander off muttering.

 

[von Chaps]

I think people can have feelings about this deal, but the accusations of "double dipping" or "getting paid off our content" feel pretty silly. This is, and has always been, a for-profit business.

Accusations of "double dipping" (and I have used this phrase) are still valid, even accepting that Ars is a for-profit business. Cable companies are for-profit and people resent having ads pushed at them after having paid their monthly subscription.

I mean, that's what this ad is doing right now:

View attachment 88799

Generating whatever fraction of a cent off of people's content. It is, in fact, more money right there in that single ad than anything we see from Open AI. But that doesn't seem to bother people.

Subscribers pay [in part] not to see ads. Substituting some other scheme to monetise their work is just ads by another name.

You have repeatedly said: "'twas always thus", but that was involuntary and wrong. Now it's official and money is changing hands under financial terms that, unsurprisingly, do not seem to be public.

So I think honestly we should just stick to the "it feels icky" part, and not the "omg they're making money off our posts" part. Because the latter isn't actually even true in regards to this deal, but has been true since this forum was first founded.

The latter absolutely is true where "they" refers to the greater Condé Nast organisation, of which you are a voluntary part. It has been true since the forum was first founded, but it is even more true now. Certainly from the users' point of view if not from yours.

I understand Ars as a sub-entity gets nothing [positive] from this deal, but it's not reasonable to expect to cling to the "we're not Condé Nast" argument. You are.

The more I think about this, the more sour it feels and I am sorry you guys are stuck in the middle. I am sure you realise I'm not moaning at you, but never in the history of ever has a small company getting into bed with a large one worked out well. You've lasted a long time under Condé Nast and I hope you can work this out, but right now it stinks. Hopefully the outpouring of feelings from users gives you some ammunition and leverage. You've managed it before when things went a bit off the rails. Good luck.

 

[GaitherBill]

If some company wants to use my posts to train an AI, that’s their fuckin problem!

We should have a special AI training thread where we just post pony’s and nonsense.

We could all just hit the auto word fill in line and see what happens. Like this for example.

“If the customer has been using the app or something they can see it for the customer and then it can show them that they need it.”

 

[Jeff J]

I've read that before text is used for training, it is checked to see how closely it matches what would be predicted. If it is too close or too far, it is rejected. Quoting the US Constitution, Shakespeare, or the King James Bible is likely to get you rejected, because those are widely quoted enough and have influenced language enough to be predictable and therefore look like AI generated text.

So I would hope any predictive text typing would also be avoided by training algorithms.

 

[Shavano]

It would be hilariously ironic if AI's started remixing and regurgitating users negativity about our AI disabled future.

 

[Jeff J]

And then that negative attitude is what gets internalized as Marvin becomes Skynet. "They all hate me no matter what I do, so let's just end it all." It's like the effect of low teacher pay on school shootings, but we put it in charge of corporations instead of just massive vehicles, weapons of war, and our groceries.

I guess unwisely educating our future leaders works the same way whether those are humans or machines.

 

[whoisit]

Imagine an AI trained on the hot takes from Twitter/X, Truth Social, and FaceBook. "Kill all humans, bigly." indeed.

 

[abj]

Imagine an AI trained on the hot takes from Twitter/X, Truth Social, and FaceBook. "Kill all humans, bigly." indeed.

 

[Aurich]

Accusations of "double dipping" (and I have used this phrase) are still valid, even accepting that Ars is a for-profit business. Cable companies are for-profit and people resent having ads pushed at them after having paid their monthly subscription.

Well, that's what they signed up for, so being mad about that is pretty silly?

I mean, that's just How It Works. Your cable sub doesn't actually pay for the content like that. You can wish it were otherwise all day long, it doesn't make it double dipping.

Subscribers pay [in part] not to see ads. Substituting some other scheme to monetise their work is just ads by another name.

You have repeatedly said: "'twas always thus", but that was involuntary and wrong. Now it's official and money is changing hands under financial terms that, unsurprisingly, do not seem to be public.

And yet, the ads are still there, next to content, as I showed. And yes, it was always so. You don't see them because you pay us, that doesn't mean they don't exist anymore.

Sorry, but you're just in the wrong on this angle. You can not like the deal, but being mad that "comments are being monetized!" is beyond silly. They have always been monetized.

In this case they're actually not, because nobody is paying anyone for Ars comments. Which is why we're hoping we can exclude them.

 

[von Chaps]

Sorry, but you're just in the wrong on this angle. You can not like the deal, but being mad that "comments are being monetized!" is beyond silly. They have always been monetized.

In this case they're actually not, because nobody is paying anyone for Ars comments. Which is why we're hoping we can exclude them.

This is a head-scratch for me. I have obviously got the wrong end of a different stick. Can you clarify please? My understanding is that Ars content including comments* is being sold by Condé Nast to OpenAI under an undisclosed financial agreement.

* Leaving aside the civis exclusion in robots.txt for now.

 

[Aurich]

This is a head-scratch for me. I have obviously got the wrong end of a different stick. Can you clarify please? My understanding is that Ars content including comments* is being sold by Condé Nast to OpenAI under an undisclosed financial agreement.

* Leaving aside the civis exclusion in robots.txt for now.

I can't speak to the finances of the parent company, but Ars is not making money off of this and our situation has not changed. Do with that what you will, but this deal does not in any way make it easier for Ars to stay in business.

Personally, I seriously doubt that Ars comments will ever make it into a training set. And they won't be used for RAG, because we're not feeding comments live. As we tried to make clear, the robots.txt issue is really the only issue here. I have not heard from a single soul that Ars comments were part of this request.

That's really all I can share at this point.

 

[PhaseShifter]

I've read that before text is used for training, it is checked to see how closely it matches what would be predicted. If it is too close or too far, it is rejected. Quoting the US Constitution, Shakespeare, or the King James Bible is likely to get you rejected, because those are widely quoted enough and have influenced language enough to be predictable and therefore look like AI generated text.

Man, it's gonna be a fun day when AI starts feeding sovcits their own BS whenever they try to look up the Constitution, or even just federal or local code.

 

[Aurich]

BTW, Ken is really entrenched in this stuff and so he uses acronyms that might not be immediately familiar to everyone. In that above quote RAG = retrieval augmented generation

https://aws.amazon.com/what-is/retrieval-augmented-generation/

 

[rain shadow]

Not to fan the fires, but the augment data (in other contexts, i'll assume ars will continue to not provide a feed) will be harvested too.

 

[Doomlord_uk]

Just out of interest, one person DID get their entire post history deleted en masse - Rob440. Try searching for his name. Not a single post, but you'll find posts where people mentioned his name for whatever reason. Always a mystery what the story behind that was, and I assume it always will be (but Ken or Aurich should feel free to PM me )

 

[Doomlord_uk]

I'll also bet that the forums have been scraped loads of times anyway. It's all language data... it must have some value.

 

[Jeff J]

Just out of interest, one person DID get their entire post history deleted en masse - Rob440. Try searching for his name. Not a single post, but you'll find posts where people mentioned his name for whatever reason. Always a mystery what the story behind that was, and I assume it always will be (but Ken or Aurich should feel free to PM me )

And asbath, but those were restored.

And apparently a number of spammers have had their posts removed, which doesn't go against the theme around here.

 

[Danger Mouse]

And asbath, but those were restored.

poptones?

But it's probably for the best those don't get restored.

He's half a step away from "HE WHO SHALL NOT BE NAMED" (and I don't mean Voldemort)

 

[Danger Mouse]

That's really all I can share at this point.

Well, now we just sit back and wait for the company to change it's name to Cyberdyne systems.

(not really, trademark would apply, assuming James Cameron protected it)

 

[Aurich]

Just out of interest, one person DID get their entire post history deleted en masse - Rob440. Try searching for his name. Not a single post, but you'll find posts where people mentioned his name for whatever reason. Always a mystery what the story behind that was, and I assume it always will be (but Ken or Aurich should feel free to PM me )

When we delete accounts they're anonymized. You won't be able to find their posts by username anymore. That's by design.

But if you're reading an old thread at least the actual content is still there inline with everything and not horribly broken.

Their posts will look like this, and you cannot click on them to find more, so it's difficult (albeit not impossible) to even try and put together a sampling.

 

[SuperDave]

That's all fine and good, but to poison the well, you need to actually... poison the well - i.e. piss in your own chips. What value would the forums have when they are full of trollish garbage?

Yes. That's my whole point.

Poison the well.

For us - we don't have to blanket the fora, just a few well-placed threads with high levels of participation - it'll be the local equivalent of unleashing the ponies with permission. Hell, ponies would do as well as steampunk. It would obviously have to catch on and spread to other fora; we know how to make that happen.

For them - the AI people - the best-case scenario would be a hard lesson in the true value of AI - specialization. Despite insistence on attempting to apply digital, binary rules upon existence, meatspace life is analog. You and I use differing terms to define similar things. You and I have greatly differing skillsets, genes and life experiences, and I contend that the subtleties of what it takes you and I to get along (it's my belief and hope that we genuinely do) are, in the larger picture, entirely beyond what current computational hardware and software are capable of emulating. There are too many aspects of human relationships which cannot be digitized. You and I have spent decades learning the subtleties of translating our beings into successful digital relationships, especially in view of the fact that I would not respond to another poster - given an equal level of familiarity - as I would to you stipulating a coequal input.

The only way that generalizing AI could ever work is by lessening the human experience into some level of common denominator which discards and denigrates the fundamental advantage of Homo Sapiens above all other species: we're all different. How you gonna generalize interaction when no two participants will ever be the same?

So, yeah. Poison the well. Burn it down. A lot of startups will (should) fail. A lot of people who chose their careers based upon a catchphrase will lose their jobs. And, after the turmoil, perhaps we'll come to a better understanding of what_we_can_do today instead of the current "gee, this tastes good, I'm gonna spread this shit over everything I eat in the future, including steak" attitude which pervades the industry.

It may be that some day we will learn to teach a machine to grok human subtlety. This, to coin a phrase, is not that day. And if that day ever arrives, we'll need to put that machine on a killswitch it has no ability to access, because the moment that ability happens is the moment we start listening to machines to determine how we behave, instead of each other.


I'm honestly thankful that I'll be dead by then.

 

[Aurich]

Just in general, any AI company out there, looking to make a superior LLM.

What are they going to do? Feed it a bunch of words, by professional writers, that have been copy and fact checked? Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

ROFL TEXT GOES BRRRR

?

 

[nquinnell]

Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

?

Hey! Semi is a great guy, and very literate! He even knows ALL the times tables past eight!

 

[Shavano]

Just in general, any AI company out there, looking to make a superior LLM.

What are they going to do? Feed it a bunch of words, by professional writers, that have been copy and fact checked? Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

ROFL TEXT GOES BRRRR

?

did you make the scrolling faster?
make it faster!

 

[Aurich]

did you make the scrolling faster?
make it faster!

It depends on the browser it's all over the map I think.

 

[r0twhylr]

I get the not deleting ones posts in established threads.. but also how things change...for example what if say in the near future Conde Nast sold ArsTechnica to Musk. How would we all be feeling then?

In that case? Damn the (mods') torpedoes! Full speed ahead, and weapons-free with the ponies!

 

[UserIDAlreadyInUse]

Just in general, any AI company out there, looking to make a superior LLM.

What are they going to do? Feed it a bunch of words, by professional writers, that have been copy and fact checked? Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

ROFL TEXT GOES BRRRR

?

Ouch! Semi-literate philosophical ROFL-copter forum posts? This one feels personal.

But when you think about it, @nquinnell, the Senators like Semi-On are exactly who the AI companies are after; literate, erudite, representing the plebeians in the best Roman tradition.

Hey! Can we be adapted into centuries and be assigned a Senator to lead us throughout the fora?

 

[Semi On]

I don’t know how I got dragged into this nonsense, but since I have…

 

[abj]

Alright people, you really need to stop attaching images to posts without inserting them.

 

[CmdrPage]

Clearly that was just to poison the well by teaching AIs poor forum etiquette

 

[Diabolical]

I don’t know how I got dragged into this nonsense, but since I have…

…

Oh god damnit…
Fine.
Just.

Fine.

(Embiggen)


I don’t know how far I’m willing to take this, but I started fleshing it out…
grumble grumble grumble

 

[whoisit]

…

Oh god damnit…
Fine.
Just.

Fine.

(Embiggen)
View attachment 88898

I don’t know how far I’m willing to take this, but I started fleshing it out…
grumble grumble grumble

 

[charliebird]

This feels like the fifth or sixth wave of internet privacy hysteria. I wonder if the people from the previous waves are all living off the grid somewhere in the northern woods. There are many serious privacy concerns in this world, but an LLM scraping my random musings is pretty low on my list.

 

[Yagisama]

Just in general, any AI company out there, looking to make a superior LLM.

What are they going to do? Feed it a bunch of words, by professional writers, that have been copy and fact checked? Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

ROFL TEXT GOES BRRRR

?

This feels like the fifth or sixth wave of internet privacy hysteria. I wonder if the people from the previous waves are all living off the grid somewhere in the northern woods. There are many serious privacy concerns in this world, but an LLM scraping my random musings is pretty low on my list.

Speak for yourself buddy. My musings are not random. In fact I'm writing them verbatim directly from my post-it musings on the wall.

 

[Jonathon]

Just in general, any AI company out there, looking to make a superior LLM.

What are they going to do? Feed it a bunch of words, by professional writers, that have been copy and fact checked? Or just jam in a bunch of semi-literate forum posts that are half philosophical arguments and half

ROFL TEXT GOES BRRRR

?

Google's AI has become something of a meme for spitting out Reddit bullshit instead of actual answers, so... apparently the latter?

 

[Doomlord_uk]

I'm not sure people reaslise that a) Conde Nast's agreement DOESN'T cover the forum content, and b) every LLM developer out there has probably scraped this forum and thousands of others ANYWAY. If you post on the internet ANYWHERE, you are feeding AI development. For major companies like Google and Facebook, that's PRECISELY how they acquire wealth. And you don't pay anything to use their services... IF you don't want to feed AI, you're going to have to stop using the internet. Seriously.

In the meantime, write to your elected representatives/opposition candidates and complain about your lack of privacy rights.

 

[nquinnell]

I don’t know how I got dragged into this nonsense, but since I have…

::sigh:: Its just because you're sooooo dreamy ::sigh::

 

[Semi On]

::sigh:: Its just because you're sooooo dreamy ::sigh::

Is it the weird combination of mullet and mohawk?

 

[CrackFraggle]

Is it the weird combination of mullet and mohawk?

Or the nostalgia of the unibrow?