Litbowl

Don't worry, they're not, given this "vision of the future" they have (which if they meant it honestly would mean shutting down the entire enterprise, because experts do not consent to their work being scraped and remixed/regurgitated period) bsky.app/profile/onsl...

Don't worry, their new vision is that experts will want to "create AI versions of themselves" for Grammarly to use!!

Ok, you're saying "experts" can choose how to participate? I think every single expert, or like 99%, chooses "you not only cannot use my name, you also cannot scrape my work or involve it in any way in your plagiariasm machines, nor can your third parties". Glad you're shutting down ChatGPT etc.

(Hope you and others are still moving forward with legal action; them stopping is more "admission of fault" than "solution", and the precedent of it could be very impactful!!)

I am surfacing the inside of the tree

And hopefully the lawsuit(s) still happen. If you like... ran an ad for Pepsi saying "Barack Obama LOVES Pepsi", stopping the ad doesn't mean you didn't violate the law and owe restitution.

That's the case for the entire last 5-plus years of AI theft. Models could've been made with licensed and public domain content. Instead, 100+ lawsuits and an unspeakable additional hatred from millions for the whole enterprise

Beyond the delightful way in which this kind of experimenting helpfully undercuts the idea that LLMs are creative or "writing", it's also fascinating to see a much more direct parroting of SOMETHING in the interplay between prompt, then "reasoning" recursive addition to prompt, then... that.

Phenomenal piece, thank you!

Re: "LLMs are not designed for verbatim playback of text", reminds me of this bizarre ongoing thing I found a couple weeks ago. bsky.app/profile/litb...

What a perfect short poem.

From Andrea Cohen's book, Everything: bookshop.org/a/862/9781945588686

#poem #booksky #writing

SLOPPELGANGER

I think it is/rumblings of it being in the works

But also in the creation of massive amounts of workslop for others to deal with it doesn't actually clearly create "efficiency" either

What a perfect short poem.

From Andrea Cohen's book, Everything: bookshop.org/a/862/9781945588686

#poem #booksky #writing

oh wild

For clarity, this is from LAST march, not this march. And it is a major failure of the celebrity class that they haven't sued these fucks out of existence, leaving it to us plebs.

Yeah go take it again in an incognito browser (it'll let you) and you can see. It's funny because best bet on what happened there is that they sent it out in their daily news email today, and that dramatically shifted the results toward human vs. people who had found it previously

Yes, entire thing is meaningless and solely meant for this idiot to feel better about his lack of any interesting/creative thought, BUT no longer the case (most readers now choosing human on all except for the plagiarized Sagan)

what a fucking loser lol

I have a really good answer for how to control it. Ban public-facing LLMs.

130 responses, I love this place.

are you SURE???

This is just blatantly illegal even without the AI part being involved at all

It's not murder if they deserve it

Thank you to the 50+ people who have responded so far. These very scientific results show that the NYT should strongly consider employing the pig for tech analysis going forward.

Lmao I actually really respect Anil's thoughts and approach for the most part even though I disagree with some of them!

C'mon guys, just ban the whole fucking thing, jesus

Oh, Mary. A perfect poem for this warm March week.

From Devotions: bit.ly/devotionsLB

#poem #booksky #writing

Is it because you thought you were choosing the person but you were actually choosing the pig's butthole?

Shit I think he found the poll...