The idea of using LLMs for ideas is so sad to me. Isn’t basically all of the fun of coming up with a new drink, the coming up with part?

This is a guess but I feel like it was around the time that most coding was done for things that weren’t explicitly “programs”, like web design CSS/Java and smartphone “apps”.

Just curious, what does “some automation” entail? I thought LLMs could only work with text, like summarize documents and that sort of thing.

This appears to be one guy’s blog who writes about the royal family, with no connection to the family itself.

He also outputs a suspiciously large amount of articles/day. Probably using AI.

It seems to be implying that SoCs will soon become so fast and efficient that they will always outperform custom builds in every category including price.

There’s an argument to be made there (not saying I agree).

Booklore was discovered to be vibe-coded and riddled with security issues. The dev shut down the project when discovered. Best avoided for now.

KOReader is not “fake epaper”, it’s an app designed to be used on ereaders which means it’s UI is high contrast. FWIW I agree it’s not ideal for an OLED phone screen, but it’s definitely not fake epaper (more like the opposite) and makes me think you might be accidentally using something else?

(Also KOReader is not for “certain” devices, it’s FOSS and installs on basically any ereader, including Kindle, Kobo and anything running android like Boox).

What CWA does is it integrates a KOreader sync server (can also be ran independently). In future updates CWA’s web reader will sync with this progress, but for now it only shows up in the UI like this:

THAT ALL SAID, if you are only using the WebUI and don’t want to wait for CWA to update their web reader, Komga is a simple app (originally designed for Manga but will work fine with books) that has a web reader that will also remember your progress (and fwiw there is a KOReader plugin in case you want to sync that progress with an epaper device in the future).

It was the UI he didn’t like so it’s not going to be much different.

At some point they are going to have to make money.

Yunohost seems the community pick these days. I also played around with CasaOS and found it very user friendly. Though development on that one seems to have stalled.

It’s also a pretty big exaggeration of what actually happened which is that it generated and posted some technically inaccurate information.

This is cool, so I could theoretically just set up a Plex/Jellyfin library and not need a youtube account?

IIRC this was an indirect response to the Unabomber’s manifesto, and the rise of the milita movement/neo luddites in the 90s.

Yeah this was my thought ever since I heard that they saw LLM “search” as the future. “Search” inasmuch as it still exists on Google is already pay-to-play, this is just changing who gets paid.

Reminds me of Elon’s comment about programmers value being based on the number of lines of code they’d written.

I agree with this. It also makes me wonder if it would even be possible for a nonprofit or government to train an image generator (for example) to produce more detailed work. The output of a corporation can only be “good enough”.

Yes thank you elaborating. I do this this situation is different than (for example) what was going on with Booklore.

Yeah and the dev didn’t say he’s switching over to a 100% vibe coded model, just that he’s ok with it as long as the bugs are manageable.

Someone once said to me “If a creative project is of high quality, the longer you look at it the more details you’ll notice. If it’s bad, the longer you look at it, the more detail you notice is missing.”

And I think that about sums it up. AI slop is programmed to be eye catching but can’t produce much detail beyond that.