SihnGaia (@Raen4657) header
SihnGaia (@Raen4657) profile picture

AI Art and Tech Hobbyist Internet Boomer Stable Diffusion Hipster I will fill your feed with arguments about AI. You've been warned

Las Vegas Nevada
Joined October 2016
Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)Photo by SihnGaia (@Raen4657)
Lots of weirdness goin on around here
Hey @X, what's going on with these shadowbans lately? While I'm sure that there are those who are upset by me defending generative AI, I've broken no rule by doing so. In fact, I've discussed things rather civilly. At worst, I call out people's poor behavior. #AIart #Shadowbans
1
125
Damn you really went and hide links to evidence you were wrong? Shameless, you guys never cease to amaze blog.andymasley.com/p/the-ai…
unavailable
2
1
2
329
So much data with sources right at their fingertips, but if they hide it they can act like it doesn't exist.
1
2
136
If art is a luxury, so is getting paid to make it. We live in a world of things that used to be luxuries, but thanks to the advancement of technology and society, are now just commonplace. It's wildly selfish to live in this world and cry once that progress comes for you too.
unavailable
1
2
7
387
These anti-AI types *love* to just block people who push back. It's wild how willing they are to self-impose their own ignorant echo chambers.
2
5
237
The responses I get from anti-AI types when I point this out is always a laugh. Always followed by an immediate block. "No correlation!" Buddy, I just showed you the correlation.😬
This is the same argument purists made against digital art. One of many actually. "Lazy" "Cheating" "Soulless" "Made by a machine".
1
303
This is absolutely not how diffusion models work in the slightest. The scraping happens before any generation is possible, as part of the training process. It learns patterns, much like people do. Like how a nose tends to fit between two eyes, or what the color red looks like
Replying to @BiZeGuo1
Again, AI is stealing from artists by scraping art, and their one sentence prompt isn’t the most precise either, as the AI just finds art online that aligns with that prompt and uses that to make the ‘art’. Tbh, they could’ve given me the prompt and that’s better than AI.
1
2
253
Those patterns are what is used to generate novel images, by chiseling them entirely out of gaussian noise without referencing any of the images it was trained on. The internet isn't even needed for this process. You can test it yourself
1
1
129
You also have no idea what kind of work went into their AI art, unless they tell you. I understand all you know about working with AI is prompts, but there's a vast array of tools and techniques to learn out there, capable of all sorts of different things.
1
89
Yes they are. LLMs are absolutely generative AI. Just like diffusion models made for images. Both are trained on vast datasets to learn patterns through gradient descent, tuning a neural network to generate novel data using those patterns. One learns from text, one from pictures.
Replying to @ecutruin @ivycomb
Just because someone uses ChatGPT doesnt mean they are pro-genAI. GenAI and LLM are not the same at all.
4
7
368
I can only respond like this because OP blocked me, but thanks for giving it some real thought. I also enjoy the process, the experience, not just the consumption. Glad you understand it's personal now though. It's a rare thing to see from the anti-AI types here on twitter.
So when I talk about it, it goes way beyond just rules, it's a way to grow personally, or worsen, too. It's not *just* producing something, it's the whole experience which gives soul to art, and I can't describe it on tweets. We agree to disagree. Have a good day.
4
2
341
Explaining how GenAI works to people who wear their hatred for the term like jewelry is always rough. They're operating from a complete lack of understanding on how any of it works to begin with, so explaining how their claims are based on misunderstandings is almost impossible.
3
1
13
484
But these people don't understand or even know about any of that because they have created their own fantastical idea of how everything works to suit their emotions. It basically makes approaching them a non-starter to begin with, but I keep trying.
2
1
6
166
Sometimes people read my explanations and learn something new. Sometimes I get that DM from a stranger appreciative of my clarifications, so I guess that makes it worth it.
1
5
119
That poster blocked me so I can only reply this way. I personally have very little issue with AI. Yes, some datacenters are built in bad places, and yes, some people use AI maliciously, But I find it all worth the benefits, both as tools for the individual, and as tools in Med
This Post is from an account that no longer exists. Learn more
1
4
491
And yes, I'm sure some of my own data made it into these models, both text and images. I genuinely hope so. I have no issue with that.
1
238
No. Neuro is literally generative AI. An LLM at her core. Trained like any other generative AI, with *vast* datasets of high quality data, scraped off the internet to teach the model patterns, which are then used to generate novel data. Just like a diffusion model for images.
unavailable
4
7
358
I know all the myths about being trained exclusively on his own data, but its just that, myth. You can't create a functional LLM of Neuro's abilities without large amounts of high quality data. vedal.ai/advice/#:~:text=Int…
1
6
202
The reality is that every ethical concern I see from anti-AI types making claims of environmental damage and data theft also apply to Neuro. github.com/Lin-Guanguo/llm-m…
2
180