Turn off/Block AI features

[JSWolf]

Originally Posted by DiapDealer

You mean other than needing to explicitly enable it to use it?

My mistake. Yes, I mean the options for enabling/disabling AI.

[JohnnyBook]

Originally Posted by Sirtel

You're presumably thinking that people will now stop reading and only ask the AI to summarize. That's highly unlikely. Those who enjoy reading (I'm guessing the vast majority of calibre users) won't stop enjoying it just because LLMs are now integrated in calibre. And if there are any calibre users who, for some reason, don't enjoy reading, adding LLM integration won't change that either.

Not books, but Movies.... After years of doing this.... I realized it changed my movie habits. Many times when a movie pops up (Old or New) that I have not seen, I will pop over to Wikipedia and read the plot summary.

I realized after doing this for awhile, that many times (>80%) that I am no longer interested in seeing the movie. Sometimes cause its: "Nah I don't want to watch that movie." or "Hmm, sounds good, but I don't want to watch it now." More or less I have spoiled the movie, and don't want to sit through it now. I have moved away from checking wiki for movie or book summaries.

The brief blurb on the back of a book, or the teaser trailer of a movie is all I want to see of a summary.

[Sirtel]

Originally Posted by JohnnyBook

Not books, but Movies.... After years of doing this.... I realized it changed my movie habits. Many times when a movie pops up (Old or New) that I have not seen, I will pop over to Wikipedia and read the plot summary.

I realized after doing this for awhile, that many times (>80%) that I am no longer interested in seeing the movie. Sometimes cause its: "Nah I don't want to watch that movie." or "Hmm, sounds good, but I don't want to watch it now." More or less I have spoiled the movie, and don't want to sit through it now. I have moved away from checking wiki for movie or book summaries.

The brief blurb on the back of a book, or the teaser trailer of a movie is all I want to see of a summary.

I would certainly never read a plot summary of a book I wanted to read or a movie I wanted to see.

[Sirtel]

Originally Posted by farmo

I too suppose that the vast majority of readers will continue reading as before. However, why should Calibre offer LLM use on books without their authors - oftenly already robbed by large LLM providers - have any saying -- authors who may sharply disagree with their creative, scientific or artistic work be flattened by LLMs?

That's ridiculous. Plot summaries of books are as old as books themselves. Now suddenly authors find that this "flattens" their work? And they didn't think so in all the thousands of years book summaries have existed?

[farmo]

Book summaries for thousands of years have been created by readers, who put human thought in the summary (and often stand with their name for it). That's different from algorithmic processing, generated with the sole purpose of enriching large cooperations. Which may, by the way, steal the original books text for adding them to their shady training sets - another thing Calibre should not support.

[Hatgirl]

Unearthing my MobileRead account for the first time in 4 years to say I won't be upgrading Calibre past 8.15 .

While we'd all love to view technology separately from the society that birthed it, we can't and we shouldn't.

The LLMs we have access to were all trained on stolen material. I will not use a tool that only exists because of harm that was done to the authors who have given me so much joy.

Calibre has been not just a wonderful piece of software, but a reminder that another way is possible - the internet can be a place of collaboration instead of exploitation. I suppose it was unlikely to remain that way forever, but I'm sad for what has been lost.

[Dal Bhat]

Originally Posted by farmo

Book summaries for thousands of years

Whut?

[Sirtel]

Originally Posted by farmo

Book summaries for thousands of years have been created by readers, who put human thought in the summary (and often stand with their name for it). That's different from algorithmic processing, generated with the sole purpose of enriching large cooperations. Which may, by the way, steal the original books text for adding them to their shady training sets - another thing Calibre should not support.

Calibre also supports the plugins for DRM removal, which may be used to upload deDRMed books to pirate sites. And a knife may be used for killing someone.

A tool is a tool. It's not good or evil on its own. It depends on the user for what purpose it's used.

As to whether book summaries are generated by a human being or a LLM, I personally don't care. I want the summary I use to be impersonal and not reflect any personal opinions of the generator (if I want a personal opinion, I'll read a review, which is another thing altogether), and a LLM summary is good enough for me in this.

[Sirtel]

Originally Posted by Hatgirl

Unearthing my MobileRead account for the first time in 4 years to say I won't be upgrading Calibre past 8.15 .

While we'd all love to view technology separately from the society that birthed it, we can't and we shouldn't.

The LLMs we have access to were all trained on stolen material. I will not use a tool that only exists because of harm that was done to the authors who have given me so much joy.

Calibre has been not just a wonderful piece of software, but a reminder that another way is possible - the internet can be a place of collaboration instead of exploitation. I suppose it was unlikely to remain that way forever, but I'm sad for what has been lost.

Hmm. Many, many tools and scientific advances were created by causing harm or even death to someone, either animal or human. While I don't necessarily condone this, stopping using everything that was created causing harm to someone would take us right back to the Stone Age.

If you think LLMs are somehow exceptional in this, I suggest you read up on the history of various inventions and discoveries.

[kurothelazycat]

Can't the AI stuff have been made an optional download, like the other plugins? That way, for people who don't want it, we don't even have to have it on our computers, while people who do want to use, can add it if they wish. It doesn't seem like something that should have been included in the main program by default.

[Sirtel]

Originally Posted by kurothelazycat

Can't the AI stuff have been made an optional download, like the other plugins? That way, for people who don't want it, we don't even have to have it on our computers, while people who do want to use, can add it if they wish. It doesn't seem like something that should have been included in the main program by default.

What part of "their code is not loaded unless you enable them" is so difficult to understand? You have nothing on your computer except for a couple of menu entries unless you configure the AI part.

[kurothelazycat]

Originally Posted by Sirtel

What part of "their code is not loaded unless you enable them" is so difficult to understand? You have nothing on your computer except for a couple of menu entries unless you configure the AI part.

The code is very much packaged in by default. Just because the "AI" is done through API calls when the setting is enabled does not mean there's nothing built into calibre. There's still all the backend components so that calibre can reach out to those AI options.

Is it wrong to not want this stuff included by default?

[aborel]

Originally Posted by kurothelazycat

The code is very much packaged in by default. Just because the "AI" is done through API calls when the setting is enabled does not mean there's nothing built into calibre. There's still all the backend components so that calibre can reach out to those AI options.

Is it wrong to not want this stuff included by default?

I don't think it is wrong, but the space footprint looks completely negligible to me (3 Python files for each supported LLM engine as far as I can tell). Assuming there are no calls to this code unless the user really decides to use the functionality, where is the problem?

I personally have no intention to use the feature, and I don't really like the idea either, but I can live with a few more kilobytes on my hard drive.

[Sirtel]

Originally Posted by aborel

I don't think it is wrong, but the space footprint looks completely negligible to me (3 Python files for each supported LLM engine as far as I can tell). Assuming there are no calls to this code unless the user really decides to use the functionality, where is the problem?

The problem is the principle of the thing, not the actual space the files take. Their mere presence, even inactive, seems to offend some folks.

Well, as Kovid said he has no plans to ditch the AI, there is always the fork linked in this thread for those who don't even want to see a menu entry buried somewhere.

Personally I don't really care one way or the other. I can live perfectly well without AI integration (although I do find some arguments posted here fairly ridiculous), but its presence doesn't bother me and I might find some use for it occasionally.

[Maverynthia]

Originally Posted by Rand Brittain

Yeah, I'm definitely dismayed to see Calibre joining the other developers putting useless and counterproductive "AI" features into the software.

I second this. Even including the options means that you have no problems with supporting LLMs and GenAI destroying the environment and taking jobs from humans. Up to and including writing books.

Also it bloats the software.