Using AI for reading

[Deskisamess]

Originally Posted by DiapDealer

Three similar sentences in every thread where the letters A and I appear together. Cumulatively, it's a freaking campaign.

Too many people are totally unaware of the true cost of AI. Bringing that info into every conversation about it is not a bad thing. At the very least it may cause the uninformed to do some of their own research about it.

[DiapDealer]

Originally Posted by Deskisamess

Too many people are totally unaware of the true cost of AI. Bringing that info into every conversation about it is not a bad thing.

Sure. As long as you agree with the premise, it's a great thing. For the people who just want to have a regular conversation about practical usage of an emerging technology, maybe not so much.

[rowe]

I find AI to be an excellent tool for book recommendations and for discussing what I'm reading, largely because the better models can draw upon a huge range of context beyond just the text itself. Things like reviews and study guides.

By setting up the assistant overlay on the power button of my android e-reader, I can easily share the current page when I encounter an unfamiliar historical event or concept and get a quick summary without breaking my flow. For this kind of query, I've found the reliability to be very high and the risk of hallucination to be minimal.

While resource consumption is a valid concern, it helps to have some perspective. The energy cost of this kind of use (including a relative share of the training cost) is much lower than other data-heavy activities people do without a second thought, like streaming video or playing games.

I use Gemini whose training includes Google's vast library of rights-owned material, including Google Books and Scholar.

[jackm8]

Originally Posted by DiapDealer

Three similar sentences in every thread where the letters A and I appear together. Cumulatively, it's a freaking campaign.

If they're contained in one post, no harm done. You've spent four posts complaining over it. That's more harming to the discussion.

[John F]

How could one get a summary of the first X chapters if one wasn't willing to use AI?

[DiapDealer]

Originally Posted by jackm8

If they're contained in one post, no harm done. You've spent four posts complaining over it. That's more harming to the discussion.

Nice try, but no.

[j.p.s]

Originally Posted by DNSB

I had to admire the article including "battling the AI's problematic tendencies, including what he[Jason Lemkin] described as "rogue changes, lies, code overwrites, and making up fake data.""

It does sound familiar from so many other attempts at using AI.

On the other hand, it shows just how human they have become.

[John F]

I'm with DiapDealer on this. Start another comprehensive thread on the evils of AI and let the people who are interested read it. Once a thread has started to Peter out, then reference the thread.

Otherwise, they are just more drive by opinions like:Google is evil, Apple is evil, COLOR is bad, everyone must use a cover, Kindle bad - Kobo good, ...

[DrNefario]

I mean, I thought this thread was going to be about getting AI to read the AI-generated books and cutting out the middle-man entirely.

[Sirtel]

Originally Posted by DrNefario

I mean, I thought this thread was going to be about getting AI to read the AI-generated books and cutting out the middle-man entirely.

Cutting out both authors and readers?

[Sirtel]

Originally Posted by Barty

Or I'll come across a character and ask is this character new or did they appear previously and I forgot.

For this, a simple search of the character name is faster.

As to the rest, I'd probably use it if AIs could be trusted to not hallucinate and make up things. Unfortunately they can't, so I find myself reluctant to use them for most things beyond the very basic. I can never be sure that their replies are correct.

[Quoth]

Originally Posted by John F

I'm with DiapDealer on this. Start another comprehensive thread on the evils of AI and let the people who are interested read it. Once a thread has started to Peter out, then reference the thread.

Otherwise, they are just more drive by opinions like:Google is evil, Apple is evil, COLOR is bad, everyone must use a cover, Kindle bad - Kobo good, ...

So you want threads to be fans posting agreement of the Original Poster, even if it has questionable content and you don't want any alternate opinions or actual facts if they contradict the OP or a fan?

That would kill any forum. It would be like comments on news articles on a site in a dictatorship.

EDIT:
It's possible to be REALLY interested in a topic but disagree with other people, either due to opinions or being better informed, or logic or simple facts. So should people be allowed to post ANYTHING and no contrary posts?

[Quoth]

Originally Posted by Sirtel

For this, a simple search of the character name is faster.

Also 100% reliable. There is useful AI that gives reliable results. LLMs and Generative AI are not reliable, by design. The main design goal is plausible-looking answers, not facts.

[Quoth]

Gutenberg is now adding "AI" generated summaries on book pages that list all the downloadable files. Still don't put source published date in metadata, and often not on the web page.

How do people find them?

[rantanplan]

Originally Posted by j.p.s

On the other hand, it shows just how human they have become.

Don’t anthropromorphize LLMs. They are not intelligent, they don’t have emotions, they don’t panic. They have been made to appear that way, for our benefit, or rather the companies benefit.