I see why this seems like a good idea. I wish they hadn’t done it and I think it’s pretty offensive. Will it end well for the Guardian? Toss-up. Will it further the public good? Unclear. - ThreadSky

harkaway.bsky.social • 7 days ago

I see why this seems like a good idea. I wish they hadn’t done it and I think it’s pretty offensive.

Will it end well for the Guardian? Toss-up. Will it further the public good? Unclear.

Comments

suw.bsky.social•7 days ago

Gotta say, the Guardian's leadership team are fucking idiots. Viner has absolutely no editorial vision, which really shows in how shoddy and poor their output is, and frankly the rest of the board ought to hang their heads in shame at this. But then, they've always been two faced - talking the...

suw.bsky.social•7 days ago

...talk about employment rights whilst keeping people on precarious long-term or zero-hours contracts with none of the benefits employees get. They don't give a shit about the people who write for them, so I'm not at all surprised that they've done this.

emetcalfe.bsky.social•6 days ago

I think this will absolutely end badly for the Guardian. I appreciate that they're likely hard up for cash but what worse way to trash their reputation for integrity and accuracy than to partner up with the plagiarism snakeoil company?

harkaway.bsky.social•6 days ago

They’re rolling the dice. If they can be the default news source for a successful ChatGPT4.5 or 5 rollout, that’s huge. That it’s more likely to be a mess may not matter. I agree, though; I think it’s a mistake.

emetcalfe.bsky.social•6 days ago

What does 'success' look like here? The default news source for AI-generated word slop? Hard to imagine either their writers or readership being thrilled with that outcome.

whatnext.bsky.social•7 days ago

harkaway.bsky.social•7 days ago

What I find absolutely fascinating, though, is that OpenAI once again has money to cut a deal with a media company, but still claims it can’t afford to pay for training data.

calcourtky.bsky.social•7 days ago

Same logic of “great company!” but can’t afford thriving wages for their employees. Both of these are failed business models if the leaders aren’t put on glossy magazine covers and given tens of millions from investors seeking power.

vyoletkyss.bsky.social•6 days ago

Failed business. AI sucks at words, it's built for numbers.

thekasey.bsky.social•7 days ago

You are forgetting it relies on humans for its training. It’s always going to have to pay for live in the moment conversation to sound real…which it’s not nor ever will be.

mole9.bsky.social•7 days ago

Also The Guardian has the money to engage in this deal while inserting Katherine Viner's begging letters on every Web page and still paying Adrian Viner (sorry, Chiles) an exorbitant sum for his "I've just seen my favourite colour" columns.
{off-topic rant ends}

mteague.bsky.social•7 days ago

Is it simply a matter of the Guardian being offered massive amounts of funding? Seems as though fear of being defunded or “missing out” on their access to this unreliable tech is what’s driving a lot of these “partnerships”.

danwhitehead.net•7 days ago

There’s definitely a gangster element to these things. “We’re gonna take what we want anyway so you might as well make a deal with us so our business looks legit.”

mteague.bsky.social•7 days ago

Yeah. Coercion under the guise of a business venture is a big part of this. Fear of being left behind, as well. In any case, people can say “NO” , not sure why Guardian hasn’t.

mayleechai.bsky.social•7 days ago

🧐

neomonsterisland.bsky.social•7 days ago

Why pay the people when you can bribe the sheriff?

mckelvie.bsky.social•7 days ago

See also: Spotify paying labels but not artists

fornbirkibeinn.bsky.social•7 days ago

There's certainly an attempt to misrepresent to lower their costs, but it's also about the relative value of different training sources - studies I've seen have generally had news sources as substantially increasing accuracy, whereas fiction is at best neutral and sometimes negative.

harkaway.bsky.social•7 days ago

Then - and I really mean this - I would be thrilled if they chose not to use fiction. But that is very much not what is happening.

fornbirkibeinn.bsky.social•7 days ago

OpenAI very much don't seem interested in running the most efficient version of a model, and unfortunately I'm not sure they consider accuracy to be that important a business objective.

ericgerhardt.com•6 days ago

The minute they open that door even a tiny crack, their business model falls apart. It's only sustainable if they're never called to account for the wholesale theft of intellectual property that the entire business is built on.

perrid13.bsky.social•7 days ago

That's kinda like arguing if you can buy a sandwich you can buy a car.

Make a moral or legal argument about stealing, fine, but using this to call them hypocrites seems completely illogical.

harkaway.bsky.social•7 days ago

We have no idea what this deal is worth or what a market rate for training data would look like (specifically because the tech cos have sought to avoid one) so I’m not sure where you’re getting that. But given the valuation of OpenAI: yeah, they can afford cars and sandwiches.

rkemb.bsky.social•7 days ago

The thing is, there _are_ useful things LLMs can do here for the Guardian that don't require ChatGPT or mass plagiarism, and a bunch of companies (including British ones!) that would happily do the work.

rkemb.bsky.social•7 days ago

I think this deal is thus far more useful to OpenAI than it is to The Guardian because it does allow laundering of their reputation without necessarily doing what will genuinely benefit the Guardian.

Comments

Posting Rules

Reply