AI written books

[JSWolf]

https://www.techradar.com/news/the-a...authored-books

The Amazon Kindle has been a real boon for self-publishing authors, but its virtual book store risks being overrun by a particularly prolific new scribe: ChatGPT.

As spotted by Reuters, there are already 200 e-books on Amazon's Kindle store that list ChatGPT as the author or co-author. But because Amazon doesn't require that authors disclose whether or not they've used AI, that's likely a huge underestimation of the number of titles that AI tools have either written or co-created.

[Hitch]

Originally Posted by JSWolf

https://www.techradar.com/news/the-a...authored-books

Yeah, BUT....

How many of those have artwork inside, or on the cover, that is what was created by and should be credited to, AI? Who's to say that even one word of any of them was AI-scribed? And that the credits aren't for the cover artwork?

Too many unknowns here. We have a handful of books out there, that have AI-crafted artwork on the cover. Honestly, I wouldn't credit ChatGPT or whatever for that, or Midjourney, etc. But I can see how Reuters in its infinite hard enthusiasm for digging up "scandals" might assume that the books were AI-assisted.

And what does "assisted" even mean? Writing prompts? Pshaw.

Hitch

[jhowell]

The content I have seen produced by ChatGPT is about average quality compared with what is available in Kindle Unlimited.

[Hitch]

Originally Posted by jhowell

The content I have seen produced by ChatGPT is about average quality compared with what is available in Kindle Unlimited.

The quality of the writing, speaking from an English 101 perspective, isn't bad at all. From a Creative Writing perspective, well...that my be a different thing altogether.

I do agree with you that the typical KU quality can be terrifying.

Hitch

[DuckieTigger]

Originally Posted by Hitch

The quality of the writing, speaking from an English 101 perspective, isn't bad at all. From a Creative Writing perspective, well...that my be a different thing altogether.

I do agree with you that the typical KU quality can be terrifying.

Hitch

So, maybe it is possible to feed a badly written story to the AI, and have it get a more sophisticated grammatical and correctly spelled version, maybe adding some random details?

Last time when I asked for a story, I asked it to go into much more details describing everything. It didn't quite turn into a Tom Clancy style, but still nicely padded.

[Quoth]

And the companies providing these tools may be sued by content owners and banned in various countries for privacy issues.
https://www.theguardian.com/technolo...a-law-breaches
Note a VPN no longer makes you anonymous. The only valid VPN use now is a secure connection with your own or office server, not to anonymise an end point.
They are all blocked in the HiTech company my son works in.

[Hitch]

Originally Posted by Quoth

And the companies providing these tools may be sued by content owners and banned in various countries for privacy issues.
https://www.theguardian.com/technolo...a-law-breaches
Note a VPN no longer makes you anonymous. The only valid VPN use now is a secure connection with your own or office server, not to anonymise an end point.
They are all blocked in the HiTech company my son works in.

Ahhh, I think that they may find it is very, very difficult to argue that ChatGPT should not be allowed to "hoover up the content," when they've ALL allowed Google, Bing, DuckDuckGo, etc. to do it since the jump.

The distinction here seems to be that you can hoover up everything you want, as long as, what, human eyes read it? But NOT if machine eyes rad it? Machine eyes still read every drop of content on the Net, by Google, et al.

It's a pretty damned fine hair.

Hitch

[Quoth]

Good time to be a top IP Lawyer?
I think like Google scanning books in libraries and everywhere else vs Internet Archive scanning books the issue is what you do with the content when you've captured it.

The privacy of people using the tools is a separate issue.

Returning extracts of newspaper sites in search is a different thing to "inventing" complete stories by mixing up complete captures.

Which leads to the third issue, that search returns links and snippets of real content and the real source; these tools deliberately obfuscate the source and "invent" content that's supposedly real but fictitious. That's potentially libel as for example when ChatGPT does give a source it might quote a real source but a "fact" that's made up from scraping elsewhere. Cue irate newspapers and companies suing OpenAI/Microsoft. Or Alphabet (Google's Bard).


So maybe not as fine a hair at it might at first seem.

[DNSB]

Originally Posted by Quoth

Cue irate newspapers and companies suing OpenAI/Microsoft. Or Alphabet (Google's Bard).

So maybe not as fine a hair at it might at first seem.

There is a story about a mayor of a small town in Australia who is suing OpenAI when ChatGPT list him as a criminal who served time.

See:
ChatGPT labelled an Australian mayor a criminal. Now he's threatening to sue

Brian Hood was wrongly implicated in a bribery and corruption scandal

[ownedbycats]

Originally Posted by DNSB

There is a story about a mayor of a small town in Australia who is suing OpenAI when ChatGPT list him as a criminal who served time.

See:
ChatGPT labelled an Australian mayor a criminal. Now he's threatening to sue

Brian Hood was wrongly implicated in a bribery and corruption scandal

Bad bot!

[Kindleing]

Originally Posted by DNSB

There is a story about a mayor of a small town in Australia who is suing OpenAI when ChatGPT list him as a criminal who served time.

ChatGPT also falsely accused a US law professor of sexual assault and created a phony Washington Post news story as "proof". In today's climate that is a good way to ruin someone's reputation overnight.

https://www.thetimes.co.uk/article/b...4f657f5b45286a

[andyh2000]

It seems like the companies producing these large language model systems should have leaned heavily on the 'generative' bit from the beginning in their publicity. Touting them as like a search engine but more chatty is coming back to bite them. If it was clear to everyone from the start that they are effectively just very sophisticated predictive text engines we might be in a better place.

Andrew

[DuckieTigger]

Originally Posted by andyh2000

It seems like the companies producing these large language model systems should have leaned heavily on the 'generative' bit from the beginning in their publicity. Touting them as like a search engine but more chatty is coming back to bite them. If it was clear to everyone from the start that they are effectively just very sophisticated predictive text engines we might be in a better place.

Andrew

Who would want that? Here is your random word salad generator. It has very sophisticated language skills, the attention span of a toddler, cannot tell truth from fiction.

[Quoth]

Originally Posted by DuckieTigger

Here is your random word salad generator. It has very sophisticated language skills, the attention span of a toddler, cannot tell truth from fiction.

Brilliant.

Great toy and replaces some politicians and marketing folk?

[ownedbycats]

Apparently, there seems to be a spambot making its rounds of Archive of Our Own and accusing various fanfics of being written by a bot. (The best part is that it occasionally accuses fics that precede ChatGPT by several years!)

https://twitter.com/AO3_Status/statu...30794292621312

Almost everyone I've talked to has gotten some, and AO3 has taken the previously unprecedented step adding an option to temporarily disable guest comments globallly.

Not sure of the point of this. People thought it was some sort of lame April Fool's joke, except that it's still happening over a week later...