If genAI is a tool, can you tell me what exactly it is that it does?

“What do you mean you were led to believe I don’t know what it does” - person who asked what it does.

That assumption?

Please tell me the piece was called “who’s afraid of RedAI”

If genAI is a tool, can you tell me what exactly it is that it does?

Just to be clear, you just learned that GenAI can be used as a tool to help someone interpret text? Well at least you learned something.

This is what I mean by hyperbole being "AI-bad"s own undoing. You know AI as tool can do things. You try use what it can do as insult, but sadly all you do is errode your own credibility. It is funny though.

I suppose it’s better you’re here. You could have been out there repeating history

The point of learning long division is so you understand it. Once you understand it, THEN you can use the calculator.

That wasn’t my experience of learning long division. Not a lot of time was spent on understanding the process. A lot of time was spent on repetition, repetition, repetion until it was wrote. The division button on the calculator was faster, does offload my thinking, but it’s easier and gave more accurate results. Using my calculator in long division class would have been considered cheating though, offloading my thinking like that.

But counter to that, I also spent a lot of time in Dreamweaver chasing the dotcom bubble that popped before I entered the job market. Even with the pop we see websites everywhere for everything.

No. The corollary of my luddite argument is that tools are tools. Attacking the tools don’t work to solve systemic problems.

If genAI is a tool, can you tell me what exactly it is that it does?

See my first comment for one example of students using it as a tool.

See, the “AI bad” people are funny. In a reply to comment about how people are actually using GenAI as a tool to achieve their goals “If genAI is a tool, can you tell me what exactly it does”? I must say, admitting to not not having a clue what genAI even does is a massive hit to your credibility on this topic. But to answer your question about what it is that generative AI does as a tool: it generates things. Pictures, videos, text, music, it generates things.

It is a shame the airways a late clogged with this nonsense instead of how the wealthy are using tools, any tools, to concentrate wealth.

Look, GenAI is a tool, use it for what you think it’s useful for (even if you think that’s nothing). Let others use it for what they think it’s useful for. But for the love of god, attack the system not the tools.

Someone just told me that Ecco the Dolphin, by virtue of not having AI, is saving the planet. There really should be a comm devoted to the things AntiAI peeps say.

I didn’t really say AI bad,

Implying GenAi gives wrong answers isn’t saying AI bad?

If you ask AI the same question twice you get 2 answers, different AIs give different responses, different prompts, different people, different geography.

That’s true of people too, and we trust them to do all sorts of things. Ask 20 people what happens after you die, how many answers are you getting? Not just that ask any technical question, ask 4 beekeepers the best way to do a thing and you’ll get 5 answers.

GenAI is a tool, if you try use it to hammer in nails you’re gonna have a bad time. Don’t try use it to hammer in nails.

It turns out wrote answers to wrote questions is something it does fairly well, and it’s still getting better. That’s good, as a society we’ve moved past wrote answers to wrote questions. We should now prepre kids for the society they are going to grow up in, one with GenAI. Critical thinking is something it does fairly poorly, critical thinking is something we do fairly poorly, let’s teach that.

Beyond academics, shitty throwaway art is something it can also do fairly well. Just want an image use GenAi, want a master piece get a human. You already do this, how many of your clothes are handmade? Used a milliner recently? The Luddites taught us a lesson, attacking looms don’t work.

I hope I was the last generation to spend hours on long division with quotes of “you won’t always have access to a calculator”. Those that go into fields where long division may be useful should learn it, the rest of us have calculators.

Nah I grew up with the “you won’t always have a calculator in your pocket” crowd.

If AI gets everything wrong then students using it to offload their thinking will get failing grades. AI getting everything wrong is a self solving problem.

But sure, attack the person, not the argument. I’m sure we’ll have a well reasoned discussion.

The AI bad crowd are really tripping up the capitalism bad crowd. I don’t know what to do about it. I don’t want them stop, they’re hilarious:

The post complaining that GenAI is used for CSAM, tools used for CSAM is bad, and the people that use those tools are bad people. The tool used to complain about AI: the internet!!! The internet, famously absent of CSAM.

The constant hyperbole is funny. Not so much the witch hunts. We can all see the post, students using AI to get through classes means AI can get their questions right. The hyperbole is funny.

Accusing anything and everything of being AI is pretty shitty though, knock it off.

But, they do distract from genuine concerns about how capitalism is using tools, any tools, to concentrate wealth. The tools are tools, it’s the capitalism that’s the problem.

I assume everyone here is wearing cloth made on a loom. The Luddites taught us that attacking the tool (the loom in that case, GenAI in this one) doesn’t work.

It’s good that students are using ai to cheat then. We won’t need to detect it as the answers are wrong.

Worked for the dotcom bubble. It blew up and we were left with corporate hellscape internet, not so/interesting independent internet, and the dark web.

AI will blow up leaving a few massive players, the Google/Facebook/etc equivalent. Some independent people doing interesting and not so interesting things. And a dark web.

I can’t vouch for it. But my understanding is that DOS is it’s own wild west of Case By Case Basis. My understanding is that if DosBox pure would be happy with the .zip you give it, Romm may or may not be. If DosBox won’t fire it up with no tinkering god help you.

I got fallout to run in a desktop instance, but it still didn’t run on Romm. Gpt suspects it’s the nested folder structure

/INTRPLAY/FALLOUT/[game stuff here including data folders]

Gpt is wrong more than it’s right though.

I’m fine with people paying for the Just-Works™.

I’ve just ventured into DosBox stuff, starting with pre installed game folders (best case scenario). Just mount the .7z into dosbox and… Wait, unarchive it, make it a .zip and… Wait, just unzip it and remake the folder structure to match the .cfg, rezip and… Wait, unzip and re-write some of the setting in the .cfg, re-zip and… Wait, mouse support on a touch screen, why is the keyboard unresponsive?

Tinkering is fun, but I just wanted to play fallout 1. Romm just added dos support. I was elated. But all my games are .7z. and the 2 I tried (FO1 and a 1-0 soccer, it was first on the list) both didn’t fire up, fallout 1 required tinkering to fire up in a DosBox instance. A fun project, I wasn’t gaming though.

I run 2 instances of pihole/unbound as lxcs on my main server and local back up, works great.

If I didn’t have the two big boxes I’d use my pi4/zero2 to run two instances of pihole/unbound.

If I didn’t have my pis, I’d run 2 instances of pihole/unbound on literally anything I could install it on.

What I’m saying is that I consider pihole/unbound to be essential infrastructure at this point. I’m also trying to say I’ve broken my only instance of pihole enough times to understand the importance of redundancy.

I use Pis as a (sort of) hardware key to get family and friends onto my Tailscale VPN. They all have pihole too. I haven’t convinced any of them to get a pi0 as a redundant box, but I’m sure they’ll learn eventually too. No doubt it’ll be my problem.

The difference between a million and a billion is about a billion. Millionaires are closer to zero than they are the ultra rich.

Cleanuparr+huntarr

Add a bunch to Prowlarr (filter by public, Torrent, Language, category) and then monitor the stats in Prowlarr to remove the ones not working out for you.

MagnetDownload has the highest number of successful grabs for me. Followed by TorrentProject2. I run a Bitmagnet instance though, and I have that as highest piority.

No you won’t, that’s where Fido sits. Fido was there before you, Fido will be there after you.

You are permitted to drink near Fido on payment of a treat, and to scritch Fido should you pass a sniff test.

When your an Atreides they just let you do it.

WAF has consistently held me down to earth. “What will that enable you to acheive that you can’t do already?” With a couple mini pcs and a rpi I’m good. I’d love shiny things, but beyond LLMs there’s not much it would enable me to achieve that I can’t already.

That, and other hobbies. I don’t want to be an amateur system admin during the summer. So all winter long, while I’m tinkering, I’m adding up how much it adds to my maintenance schedule.

Honestly, I wouldn’t.

I only run it this way because a VPS had 0 WAF, and I’m terrified of opening ports. VPS is the well trodden ground, there’s tonnes of guides. Mine’s a hack job borne of necessity, it works though, and I am proud of what I cobbled together.

It was my first time solving my own problems. I had my meager skill set, a basic idea of what I wanted, some vague notion of how I was going to achieve it, and a thick forehead to smash against the problem till it gave way for me.

I am going to keep running it this way though. To access my server you need to HAVE a relay rPi, and you need to KNOW a password. That’s two authentication factors right there, just built in.

I use tailscale for my non-tech family.

I run a rPi with tailscale, pihole and nginx on it in their house. They connect to the their WiFi, get adblocking for free. They go to “http://homarr.sever/” pihole captures the request, sends it to nginx which reverse proxies to a homarr LXC on my server. From there they can click links to the services which are at “https://service/.######.xyz”. Again, pihole captures the request, sends it to nginx which reverse proxies it over Tailscale to the appropriate LXC.

One poor soul runs a mini pc with 2 mirrored ssds attached, it runs everything above plus Syncthing. They have the privilege of running the remote back up for the server.

For apps on their phone, I intend to set their phone up with Tailscale and then just have the app go to “http://dockge:1337/”… Just as soon as I learn to write the access controls to allow admins to access everything, users to access services, and services to access nothing. I just looked and there’s a gui now so I could maybe do it this winter.

There should be both. Minimal config + gui options for people just getting into the hobby, or just want the thing. And a more open option for people who hit the limits of the first, or to do interesting shit, or to repeatably build a thing.

I go back and forth on my server. During summer I wish it was all Docker YAMLs so I can press “update” in Dockge and then enjoy the weather.

But, I also do non-typical things. Users have a rPi in their house that captures requests and routes them through Tailscale to my server for remote access without a VPS or opening ports.

I’m not too technical so I often struggle setting things up, and documentation can be less than helpful at times, sometimes I really wished there was a gui or wizard, but it’s doable.