|
|
#1 |
|
Wizard
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 2,028
Karma: 36307910
Join Date: Sep 2017
Device: PW3, Galaxy Tab A9+, Galaxy S25FE
|
Using A.I. for eBook summaries
I am not an A.I. user and I don't know much about its capabilities. I was wondering about A.I. and eBooks. Has it advanced to the point where you can give it an EPUB and it will come up with a plot summary of the book (I'm talking about fiction books)? That would be useful to me when going to read a new book in a series where I haven't read earlier books in the series in a long time. In other words, summarize books I've already read so I can catch myself up on important details before continuing the rest of the series.
If this kind of A.I. summary is possible today and it's decent enough quality to actually be useful (doesn't miss important stuff and doesn't hallucinate stuff that didn't happen in the book), how do you access it, what does it cost, pluses/minuses of using it, any other details about it? |
|
|
|
|
|
#2 |
|
Bibliophagist
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 51,823
Karma: 180002962
Join Date: Jul 2010
Location: Vancouver
Device: Kobo Sage, Libra Colour, Lenovo M8 FHD, Paperwhite 4, Tolino epos
|
One acquaintance of mine was trying to use AI to generate summaries. It wasn't that hard to set up but the main issue was that about 20% of the time, the AI hallucinated while creating the summary. This ended up requiring a human who was familiar with the books had to review each summary to remove the crud. They mentioned one example where the AI created summaries of the books in the Chronicle of Narnia had several with Aslan as the villain.
The project was finally dropped since the need to review the summaries made the process take more man hours than having the humans write the summaries. |
|
|
|
| Advert | |
|
|
|
|
#3 | |
|
Resident Curmudgeon
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 83,294
Karma: 153646249
Join Date: Nov 2006
Location: Roslindale, Massachusetts
Device: Kobo Libra 2, Kobo Aura H2O, PRS-650, PRS-T1, nook STR, PW3
|
Quote:
|
|
|
|
|
|
|
#4 | |
|
Custom User Title
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 11,783
Karma: 83105959
Join Date: Oct 2018
Location: Canada
Device: Kobo Libra H2O, formerly Aura HD
|
Quote:
![]() (source)
|
|
|
|
|
|
|
#5 | |
|
want to learn what I want
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 2,180
Karma: 7966751
Join Date: Sep 2020
Device: none
|
Quote:
I found it essential to have an account at Open Router - https://openrouter.ai/ . They offer a wide range of free and surprisingly capable models, and it's also free to get started. You just create an API key and paste it in calibre, that's it. I plan to provide detailed instructions when plugin is released. |
|
|
|
|
| Advert | |
|
|
|
|
#7 |
|
Wizard
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 2,028
Karma: 36307910
Join Date: Sep 2017
Device: PW3, Galaxy Tab A9+, Galaxy S25FE
|
|
|
|
|
|
|
#8 |
|
Still reading
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 15,456
Karma: 114319649
Join Date: Jun 2017
Location: Ireland
Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper
|
No, AI only shortens the source. No context or understanding to be able to summarise. It may sometimes use a scraped (stolen) summary or get it nearly right by accident. It will mostly even ignore a summary in the source.
Search (without AI) for actual real tests. It's a lie helping to destroy the environment. Note a blurb, synopsis, plot summary and regular summary are different things and some only appropriate to a novel. Also a blurb ideally has no major spoilers and the others should have. My own testing on a book I know resulted in a summary unrelated at all to the book. Want a summary? Use search for a real one, employ someone or do it yourself. The LLM / Generative AI will never save time. They will not improve. Mathematically proven that they get worse and eventually collapse if fed their own output, which is now happening because the makers won't pay for humans to curate the content they scrape. |
|
|
|
|
|
#9 |
|
Groupie
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 199
Karma: 5500000
Join Date: Jul 2019
Device: Viwoods Reader, Mira, Paperlike Colour
|
Understanding often isn't required for functional utility. Of course these models don't think, they're tools for pattern recognition and information retrieval. The risk of hallucination for this kind of task is low with frontier models and the output quality is usually high.
Last edited by rowe; 05-07-2026 at 06:11 AM. |
|
|
|
|
|
#10 |
|
Still reading
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 15,456
Karma: 114319649
Join Date: Jun 2017
Location: Ireland
Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper
|
Zero recognition. It's matching. Computer pattern matching works well but doesn't help with summaries.
The output quality is garbage, proven, unless the input is copies from somewhere. There is no hallucination, only incorrect output which is high unless the summary is in the "training" data. |
|
|
|
|
|
#11 |
|
Grand Sorcerer
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 29,426
Karma: 211545324
Join Date: Jan 2010
Device: Nexus 7, Kindle Fire HD
|
And here we go again...
![]() So it's the people who are satisfied with the results they're getting from AI summaries who are hallucinating? |
|
|
|
|
|
#12 |
|
Groupie
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 199
Karma: 5500000
Join Date: Jul 2019
Device: Viwoods Reader, Mira, Paperlike Colour
|
Saying "pattern matching doesn't help with summaries" ignores what summarisation actually is. Identifying narrative structures, character arcs, and key plot points is literally just advanced pattern recognition. Models don't copy pre-written summaries verbatim. They construct responses based on the weights of all their relevant sources.
If the book itself is in the training data, the model can generate a fresh, accurate summary without ever having seen a pre-written one to copy from. And if the book isn't in the training data, modern AI searches the web to fetch the necessary plot details and synthesises them on the fly. The idea that they only output garbage unless they are copying a scraped summary is completely outdated. |
|
|
|
|
|
#13 |
|
Zealot
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 143
Karma: 20456
Join Date: Apr 2014
Device: KPW1
|
I once asked AI to recommend 5 thrillers which critics praised but had low sales.
It gave me the list, including detailed plot summaries, and all sounded interesting. Except 3 of the 5 didn't exist. Relying upon AI clearly isn't a good idea, whether you're a reader asking it about books or the Pentagon giving it autonomy to kill. |
|
|
|
|
|
#14 |
|
Wizard
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 1,702
Karma: 3177153
Join Date: Dec 2010
Location: Jianghu
Device: PW1, PW5, iPhone 13 Pro, iPad Pro 2021, iPad Air 2025
|
|
|
|
|
|
|
#15 |
|
Grand Sorcerer
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Posts: 8,198
Karma: 72583667
Join Date: Feb 2009
Device: Kobo Clara 2E
|
|
|
|
|
![]() |
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Metadata: summaries not retrieved | Doctor13 | Library Management | 6 | 11-03-2022 12:33 PM |
| Problem with book summaries | agirlnamedfia | Calibre | 11 | 05-11-2019 08:12 AM |
| Book summaries on the Kindle? | Uthred | Amazon Kindle | 20 | 03-16-2013 06:34 AM |
| Exporting Metadata Summaries? | carld | Calibre | 3 | 08-08-2010 05:21 PM |
| Summaries of books available | RandmTask | Reading Recommendations | 3 | 08-24-2009 09:19 AM |