I just received an email from Cambridge UP asking for permission for them to include "Mexico's Cold War" in a new generative AI licensing agreement. Anyone else get this message? Any advice? I'm leaning "no" because I hate AI, but maybe this could help me retain attribution ability...? 🗃️ - ThreadSky

renatakeller.bsky.social • 79 days ago

I just received an email from Cambridge UP asking for permission for them to include "Mexico's Cold War" in a new generative AI licensing agreement. Anyone else get this message? Any advice? I'm leaning "no" because I hate AI, but maybe this could help me retain attribution ability...? 🗃️

Comments

mikeneagle.bsky.social•79 days ago

Got one too for "America's Forgotten Colony." Will keep tabs on this thread to get some ideas. Think I'm leaning the same as you on this issue.

mikeneagle.bsky.social•79 days ago

Adding @dgershenowitz.bsky.social here to see if she has any suggestions or insights.

dgershenowitz.bsky.social•79 days ago

I've not followed this too closely but Camb has been highlighted as having very fair treatment regarding authors with its AI policy. I'd get your questions in order and ask your editor. [or whoever became your editor when I left]

rjhammond215.bsky.social•79 days ago

I got one of these too and am also leaning no. Interestingly it didn’t seem to offer a way of saying no beyond just not signing it, as far as I can tell. I’ll keep a watch on this thread because I’d be very interested to see what others are thinking. Still not fully decided

borderlander62.bsky.social•79 days ago

I sent them a hard no re my book. Do not feed the AI monstrosity.

renatakeller.bsky.social•79 days ago

That's how I'm leaning. I don't know if it's a pointless fight, though...

borderlander62.bsky.social•79 days ago

We may not win, but why acquiesce?

skyquake.bsky.social•79 days ago

Plus we can say with some certainty at this point generative AI will never be artificial general intelligence.

I’m not saying we’ll never get there but they appear to be asking a lot of people here to feed their books into a gibberish machine.

https://medium.com/@colin.fraser/hallucinations-errors-and-dreams-c281a66f3c35

borderlander62.bsky.social•79 days ago

This was a good essay. https://undark.org/2023/04/06/chatgpt-isnt-hallucinating-its-bullshitting/

skyquake.bsky.social•78 days ago

I think the other article is one of the best I’ve seen on why we can’t expect them to ever stop bullshitting, but this article is /definitely/ correct that is the word with the semantic valence we’re looking for for randomly chucking words out in response to a prompt.

skyquake.bsky.social•79 days ago

I actually think LLM are fantastic when applied to things they’re actually good at like looking at radiology images to sus out tiny hints of cancerous patterns the doctor can check up on.

This is tragically not going to benefit your average shareholder.

renatakeller.bsky.social•79 days ago

Yes--"gibberish machine" is a perfect description

amyjpetty.bsky.social•79 days ago

Yeah, don't just go with "yes," for no other reason than believing it's pointless to resist!

arielron.bsky.social•79 days ago

Were there any conditions? Do you get royalties?

renatakeller.bsky.social•79 days ago

They said "20 percent royalties" but didn't explain beyond that. Like... 20 percent of what? My book's price? How often?

ffreff.bsky.social•79 days ago

From what I understood the company will use the books to train the AI and will pay Cambridge a fee for it. Authors of the books used would share 20% of this fee. It would be a one time thing.

ffreff.bsky.social•79 days ago

Probably the fee will be negotiated by Cambridge based on how many books are authorized by the authors to be used for training. The amended contract would also allow for new deals to be closed with different AI companies, or new iterations of AI systems.

ffreff.bsky.social•79 days ago

The contract I received says something like they would only close deals with companies committed to systems that would make correct references to the works used.

renatakeller.bsky.social•79 days ago

Yes, I like the attention to attribution--that's one of the few aspects of the whole thing that makes me even consider it

caiobarbo.bsky.social•79 days ago

It sounds as a bad deal.

christinekooi.bsky.social•79 days ago

I did, and I said no. But then I'm a 20th-century fossil who wants nothing to do with AI, no matter how many royalties CUP promises. I also found the email alarmingly vague on exactly what the contracted AI companies would do with my book.

renatakeller.bsky.social•79 days ago

Totally! Fossils unite!

cmcknichols.bsky.social•79 days ago

Seems best to opt out of all the ways AI is using "our" content, whenever possible, until a lot more gets figured out about how AI is and will use what it learns/can crawl/etc. as data science becomes increasingly industrial and AI is (increasingly?) dependent on the data it can learn from + use...

renatakeller.bsky.social•79 days ago

Yeah, there are so many unknowns about the whole situation

Comments

Posting Rules

Reply