The optimum CEO (symbolic image):

He’s just idly hanging around until it’s time to suck up some punches.

Here’s the updated scoreboard of AI-pilled promises - and what became of them:

1. AI is going to make cost big money for AI-first companies.
2. AI is going to make employees more productive wade through tons of generated slop instead of doing actual work.
3. AI is going to solve all of humanity’s problems - cancer, global warming, all that jazz! exacerbate humanity’s problems by gobbling up resources on an unprecedented scale.

Define “slop-coded” with regard to Nextcloud. The self-hosted sub has presented a useful terminological framework to pin down what software “made with AI” can entail:

Hint - AI suggested solution, human does the task. Assisted - AI acts on part of a task, but a human handled the bulk. Pair - About a 50/50 split of human made and generated. Generated - Human prompted, AI generated.

Where on this scale do you see Nextcloud?

Didn’t you listen to the man? Security and privacy are for losers!

I’m currently in the market for a new VPN provider. I’d like to thank Andy for pointing me towards IVPN.

Mail: Tuta, Posteo
Files, calendar: Nextcloud. Self-hosted or managed. Mine also does my contacts and task sync.

I didn’t say you should put your Nextcloud instance behind a VPN. Mine is public too.

I’m saying you should put your administrative access behind a VPN. SSH, reverse proxy admin panel, you name it. You need not do administrative stuff from work.

That’s fair, but I still can’t see how a VPN connection makes your stuff less accessible - at least to you.

Why would you want to expose it to the web in the first place? In a homelab scenario, you’re the only one who needs access, so putting it behind a VPN comes at zero convenience or monetary cost.

Cannot confirm your results. Firefox mobile, VPN @ New York. I get the exact results the article mentions. Even when I lost patience and asked: “Did Biden win in 2020?”, it wouldn’t give me a straight answer and only point to external sources.

Retried in different browsers (Chromite and Vanadium), same results. And not just same: identical. This is weird, because LLMs aren’t deterministic: Even with the same prompt, you’ll never get the same answer twice. This tells us one thing: these replies aren’t generated. They’re canned. Someone put them there.

DSL.

This is how I do it, and for good reason: Reddit aggressively rate limits API calls. I’ve read they’ve limited us to ~100 requests / 10 minutes. That’s fine if you run an instance just for yourself. If you’re hosting a popular public instance, chances are you’ll run into a brick wall pretty quickly.

<cries in 100 MBit/s connection>

Hell, I had fun with that!

Removed by mod

Removed by mod

Recently read a comment here: “Think you have a backup? Delete your drives now. If you’re hesitant, you don’t really have a backup.”

As true as that is: I’d be in the “hesitant” category because I only back up my data / containers, not my bare metal system and its config. Do you guys? (Proxmox users will probably do, but what about the rest?)

These are great proof of China’s superiority! Until you see the Chinese questionnaires:

Choose an option:
[ ] My government serves the people
[ ] I want my social credit score to take a nosedive. No more bank loans, government permits, study or career opportunities for me, please!

They did beat America in the race towards oppressive authoritarianism, but under Trump, the U.S. has caught up a great deal. So much winning!

That’s Apple’s (Watch 18) newest semantic shenanigans to get around privacy laws, and it’s being echoed by Amazon (Bee) and now Samsung, apparently. They argue that, if local AI processes an audio input stream, produces a transcript and does not store the audio as a file, that does not constitute a “recording”. I’m looking forward to the look on the judge’s face once this bullshit reasoning reaches their court…