Is Ai taking over the platform?

0:00- Intro
1:33 - ai explained
6:08 - ai and social media
8:48 - wide use of ai
11:57 - generative ai will not replace us

Sources:
Ai Information and Defintions

Amount of Content Out There

Jellysmack and AI Editing

Generative ai is a divisive and unpopular tool among artists. The rise and growing popularity of AI is undeniable, and the spread among creatives we love might be more than we first thought. So what is the truth behind the current uses of generative, narrow, and all Ai on social media by influencers?

[juice-lover-]

i dont think there is an issue with ai being used as a tool to help people create, but there is an issue when it's being used to diminish human creativity, or even worse, steal from human creation. there are defiantly lots of great applications for ai and you covered the topic well :D great video!

[berneymark]

I feel like you’re conflating AI and generative AI, which as you mentioned are different things. We have been using AI for well over a decade. The “algorithmic AI” we have grown up with such as Google Search is not what anybody is arguing against. It is only the generative AI that is taking jobs from people and many are stealing copyrighted content.

And why should anyone care that Billy has an idea but can’t implement and wants AI to for him. This is where Billy either learns the skill or pays someone to do it. There shouldn’t be an easy way to sidestep developers’ and artists’ hard work.

[UsurperUrsus]

I wish there was a browser extension that could read transcripts for/ listen to videos that users flag as possible AI content farms and create a database of channels that do it. so tired of the amount of slop content

[sercascade]

even before language models were popular people still believed in "dead internet theory"
it is estimated that half or more of the internet is bots

[himbosuplex]

It may not replace influencers but its already replacing non-influencer creatives and I don't think it can be any use as a tool to *aid* people until there are hard guidelines and rules in place for how it should be used in the workplace. It's easy to say "it won't replace us", and maybe it won't, in the long term, but I've already lost work to it and people don't really seem to give two shits about that. Because corporations want free labor, and if they can use AIGC to make billboard graphics for a video game with gibberish texts, or interstitials for their movie sets, or book covers, they will, and it's human workers that get fucked in the process.

I feel like until we get some guidelines and rules in place to account for the human cost of AI (like mass scraping people's work without their consent to fuel stable diffusion, or midjourney, or large language models and then replacing human works with that) nobody will ever think bigger, and it sucks. It sucks because there's so much potential for it as a tool to help create, and you see it in algorithms that can remove backgrounds from photos or videos that someone used to have to do by hand, or generating color palettes from images so artists can keep color match. Things that enhance the lives of creatives instead of trying to cut them out entirely.

Hopefully any of that makes sense, I'm tired haha.

[smartduck]

AI is used everywhere and has been used everywhere even in smartphones for well over a decade it's just it wasn't labeled AI even though it wasn't labeled as such sadly most people didn't see this development coming because of hype along the development of AI voice recognition is AI any voice to text software is AI we have been mostly limited by factors such as processing power storage and memory access speed and multi linear computational power

[smartduck]

When I was an editor for a company they had me use AI and then replaced me with it
I don't think it's all bad not surprised the video world is starting to move in that direction they are usually the first to move into the new territory and companies want to cut as much spending as possible

[oliv]

why is d'angelo wallace in the thumbnail when he was never mentioned in the video?

[smartduck]

Generative AI such as image generation AI in general is actually a identification AI which is capable of identifying objects it has been trained to identify many many many many many many many many many different objects and people places and things it's fed billions of billions of images with names for them and tags then relations usually a short description describing what the image looks like there is usually a second program which generates noise it's a Neral network what is the certain amount of parameters which is able to control each point when it's going through training the AI will have it keep generating thousands upon thousands of low resolution gradients telling the neral network turn on what it has got wrong basically just saying I don't identify the input in this noise usually these are gradient noises what are somewhat easier to generate once it's gone through its basic training and is able to reproduce objects that are recognizable by the image recognition software then the AI is fixed in place if you put two words you want to see the neural network will run and those keywords it will remember from when it had to recreate things that were recognizable basically following those neural paths that were created for those key words leaving you with a low quality image the next step is destroying that image step by step destroying that image a noise algorithm will absolutely destroy the image almost and then a reinforced learning denoising software will run removing the noise and making the image detectable with the objects using the keywords to denoise then the noise algorithm destroys the image again and the process is repeated step by step what was usually just an outline or spots or a very blurry image becomes clear and the longer it goes through these steps the more refined to a certain point because there is a limit to other iteration steps with less improvement usually bottlenecked by the denoising AI sometimes it will switch to a different AI which has been trained on larger images so it will be destroyed and then upscale to a larger image and then broken again this time less and this other software will go off of iterations after that doing layering steps 

There are other image generators which destroy an existing image then have a denoising AI reconstruct the destroyed image doing this over and over again and comparing it with the original they also include keywords and a description of the image as well and the denoising algorithm uses that as a guide it does this with billions of images people might make fun of it because they can see his AI but after the AI is able to sex images that have been destroyed it does in layers destroying them more and more and more and more until they're just noise and then the system is fixed and then if you type in a keyword and then another keyword an AI generates random noise and then tell the a I pay this image is this and this is the guide basically it uses your text as what it is supposed to see so it will generate an image from paranoids randomly generated and then there's different ways to create noise and you can also create other AIs which are able to create noise which crates composition almost like a guideline which helps the denoising AI this is also a step-by-step process 

I forgot to mention whenever the denoising AI or the other noise AI generate an image that is false in training then parameters are adjusted and there's usually lots and lots of noise and denoising AI running at once the most effective ones are the closest to the desired image gets to move on the first day I can actually can have a step whare it will ask for a combination of different keywords and check with the image recognition AI to see if the multiple keywords were retained after the image software is fixed there is no data set of images that was just for the trainer AI and recognition AI