Alexey Skobkin

@lnkr Это может быть эффект чисто на твоём провайдере.
@vlnst

@vlnst Только сейчас?
Давно ж уже все нормальные хостинги блокируют.

@IzzyOnDroid

But if so, why not name it? Must be something with "why make it transparent? Who will look at the permissions anyway?"

Yes, I think marketing people who write (or ask LLM's to) descriptions for these apps think that. And honestly for most people they somewhat right.

But yes, I'd prefer transparency myself too.

I don't assume malice here, especially not by the banks.

Well... Maybe not in your country, but I wouldn't discount that possibility...

@IzzyOnDroid

What for, I wonder?

Given my user experience with modern Russian banking apps, if we consider their intention benign, then it could be for in-app support calls or smart voice assistant (yes, modern bank apps here have assistants for a long time).

But these days you can'be sure about anything with corporate apps 🤷

@rayslava Всегда подозревал, что ты инопланетянин, а теперь когда увидел, что у тебя два пальца на ногах - всё стало ясно ☝️

@SNB
I understand, but I don't want to login/register until I understand completely what's this about, what are conditions, etc. Shoving a login form to my face isn't a very good way to catch my interest 🤔
Thanks for the documentation link though.
@hetzner

@hetzner Link requires authentication.

@Revertron Для 2.6B - не так и плохо 🙃
@news

The audacity of this bitch!

Hermes пытается меня газлайтить приписывая мне свои похождения 🤡

#pic #fun #LLM #agents #Hermes #security #gaslighting #log

@rayslava
Не объединимся - не разосрёмся!
@E23

@ToonLink I agree again. Although I wouldn't put the blame on technology myself.
From my PoV this does not contradict in any way with the fact that I don't like situations like you or David describe. I just don't want to look for a scapegoat. I want a long-term solution.

@Revertron
В нашем случае у всех троих писал бы агент 🙃
@rayslava @E23

@ToonLink Agreed 🤝

My initial reply was about that too. Although I could've expressed it better (not sure if I wanted to). People are losing their minds around this topic in a way that even David receives such comments.
I can't agree with David on everything, but I don't think the situation in OP is OK. But I also don't think that anything but time and civil discussions can help with that.

@rayslava Стадия планирования и брейншторма бывает куда интересней реализации.

Надо объединиться с тобой и Женей (который свою RPG на LLM давно пилил) вместе, запустить проект, пилить его в разных направлениях как лебедь, рак и щука, разосраться и продолжить пилить своё 🎉

@E23

@ToonLink

Here's an unfinished painting I did

Nice painting. Not ideal, yes, but nothing is.

I can't agree almost on everything except that delegating such tasks to the model will not help you learn to so that yourself. Which is objectively true.
That's a very good, legitimate and respectable reason not to use models in your work now.

To me this "it would destroy" argument makes no sense though. But it's your art and I have no right (and I see no point) to say how to do it.

I'm an artist. Yes, I understand the concept of making art.

That's not what I asked, but ok.

It's not debatable, it's already legal fact in the US.

Yes, it is. Not all the world is US. US isn't always right.

Also this is another matter. There are a lot of dumb laws in the US. And in other countries too. This doesn't mean that anyone should just agree with it and behave like it's the law of physics. And even more, it doesn't mean that laws should not be debatable.

For example there's a famous section 1201 of DMCA which basically tells you that in some cases when you want to keep owning what you have bought, you may go to jail. Something tells me that you'll probably agree that this law should be reformed or abolished entirely.

So yes, I think that some laws regarding models usage that are appearing now are being made by populists and harmful to the humanity. Like section 1201 of DMCA.

And just in case, I wasn't saying anything about legality of the matter or about business decisions of clients. Those are completely different parts of the topic and came not from "AI" itself, but from how people are reacting to it. And I won't overstate if I say that people often overreact to things (someone wished me death in this thread 🤷).

Also your arguments are much closer to me that of anyone else who I saw in such discussions. Because they're basically not "anti-AI, go die techbro", but "I won't use AI because X, Y and Z". And it's completely fine with me. I can relate to that, respect and support that.

@ToonLink

Well, no. I'd rather spend the effort making something that's actually mine. It gives me no joy to have a machine output the artwork for me. And then I have an image that isn't any different from the rest of the machine-made stuff.

Figures. Because if you had tried, you would've understood how much of what you wrote is wrong.

Can you honestly say (at least to yourself) that you understand a modern artist pipeline and how models could be integrated into it? Do you know what inpainting is? Do you know that models are often used to draw a draft over a pose sketches drawn by the artist manually? Do you know how exactly models are applied on different stages of production?

Just in case, I'm not an artist myself, I just happen to know a couple of people who draw and use "AI" models in their work. So you shouldn't take my words as being completely accurate.
I usually use simplest options like ChatGPT or locally hosted Invoke with Flux/StableDiffusion to do something non-complex for myself and don't really care if someone would call that an art or not.

What I worry about are people losing contracts because they have to prove that they actually made it.

This is a good point, but for me it's again a human problem, not an "AI" problem. It's basically an "ad hominem". If I saw a good work I like, I wouldn't think if the author used a model or not. I'd pay them anyway (if price was in my budget).
Making an author do everything manually on every stage just for the sake of my ego is a strange concept to me. Unless I pay them an enormous amount of money, of course.

I'm actually not worried about my livelihood, because people see value in my work.

That's good. And I honestly wish you having reasonable people around you who won't be questioning your work for strange reasons regardless of how exactly you do it.

Clients won't work with AI because it's uncopyrightable.

That's debatable and yet another thing that we as a humanity should work on. And mass hysteria and misunderstanding doesn't help here at all.
Unfortunately, knowing people, I don't think the problem would be solved soon. Such problems should be approached with a cool head and currently isn't a time when people can do that.

@radioclash

@ToonLink
Well, you wouldn't know until you try doing something more complicated than asking ChatGPT to draw you a picture. But then you'll probably feel an urge to rationalize something like "those hours of work are not worthy" or "using AI makes it lose it's value", or something else.
It'd be very close to any other "you're doing it wrong" debate in the field where there's no objective "right" or "wrong", but people still cling to their old ways and ostracize anyone who does something else.
At this stage it makes no sense to argue with such a person because it's already not just a thought, but a defensive mechanism.
Well, almost no sense. I usually still do that to some extent because I don't want people around to think that it's the only one possible opinion.
@radioclash

@Alex0007 You're lucky then 😀

@Alex0007 It seems like harness affects that significantly.
MiniMax-M3 is not a brightest model, but in Claude Code it uses native tools efficiently. In Hermes though I regularly see stuff like you're showing and model behavior sometimes feels lobotomized.

@ToonLink

This is what happens when people see art as a product to be consumed and not something human with a piece of the artist's soul put into every brush stroke.

This is a false dichotomy.
You don't know how much effort and time people who use models in their art put into it.

If you see "AI"-assisted art as "they just write a prompt and get an image to post", then you don't know how the process works.

If we're talking only about simple pictures people like me can use, then yes. I'm no artist. And I don't really care how would you classify it. Although even I can spend several hours to work on a preview for my podcast episode. So if you're thinking that every generated image is effortless and thoughtless, then you're also wrong. But in this case it your loss.

@radioclash