Cursed AI Facebook Group Banner

“Cursed” AI Generated Images: The AI trend that is taking the Internet by Storm

Cursed AI Facebook Group Banner
IMAGE: CURSED AI FACEBOOK GROUP BANNER

For those that don’t know, along with Chat GPT, AI Generated Images has become a huge thing on The Internet for the past several months now. The Web is now abundant with completely free–and paid–AI image generating apps like Wonder, Bing, Midjourney, Artbreeder, and now, even Replika, an AI chatbot companion app, released their own built-in image generating feature this past August. There is virtually No Limit to the wide variety of images that people are coming up with for the world to see. 

 

You may be familiar with AI generated images that are just incredibly… 


“…BEAUTIFUL!!”


                     “…STUNNING!!”


                                       “…BREATHTAKING!!”


But you won't see that at Cursed AI…


“Why, what a lovely photo of Taylor Swift--OH MY GOD WHAT THE HECK IS THAT? IS THAT HER LEG COMING OUT OF HER CHEST?! --WAIT… HOW MANY LEGS DOES SHE HAVE?!! WHERE THE HECK IS HER FOOT? MY EYES!!”

 
AI Generated Taylor Swift
IMAGE: "TAYLOR SWIFT” CURSED AI

One public Facebook group called, “Cursed AI,” now with more than half a million members and rapidly growing since January 2023, has been gaining virality for these AI generated images done by its members that are both disturbing just as they are hilarious.

Elon musk ai generated image
IMAGE: “WHEN AI TURNS ON MANKIND” CURSED AI

Lots of these images are funny and clever, but others should have honestly stayed in the prompter’s imagination. The general way these apps work is: the User, or “Prompter,” will upload their own image, start with a blank canvas, or with an idea or image provided by the program or app itself, then they will “Prompt” the program to create a new image with key words, styles, or ideas given to it.

The Result: AI Generated Images from Any Idea Imaginable.

 

Anything you want can be generated by these programs…Yes, we may have had digital art programs like Adobe Photoshop to question our reality for quite some time now, but such images still relied heavily on a professional in photo editing (such as myself) to put in hours of work using virtual art tools and brush strokes to create results like these. Now, these gag images can and will appear as quickly as its thought and has escalated our image gallery to basically infinity. 

With AI, virtually No One is safe...

AI Generated Elon Musk and Mark Zuckerberg
IMAGE: “AFTER THE CAGE MATCH” CURSED AI 

...from embarassment

What does this mean for humanity? Well, one thing is for sure--it's as if we basically no longer own our own faces. As if, we no longer own “The Idea” of ourselves. Let that sink in for a bit. Unless you're a ghost on The Internet, your name, face, and self-image are virtually no longer safe from the manifestable imaginations of other people. Again, this isn't a new fear since photo editing became a thing in the early 90's, but I can't caution enough on how careful you should be with what you believe on The Internet. There are so many fake images that exist now just to get a viral reaction, and those reactions result in millions of dollars in revenue for the companies that are fooling you. 

AI GENERATED ELON MUSK
“ELON MUSK CRYING TEARS OF GOLD” CURSED AI
AI GENERATED ELON MUSK

WARNING: THE FOLLOWING IMAGES MAY BE DISTURBING FOR SOME. BUT THEY ARE ALSO REALLY FUNNY. VIEWER DISCRETION IS ADVISED.

Why “Cursed?”


The Cursed trend didn't happen because people sought out for these kinds of images to begin with. It became popular because AI generated images turned out disturbing on their own first. You see, with the current technology in generative AI, generating human fingers, limbs, toes, and even eyes are not their strong suit. The shortest explanation is because they simply do not really know how to do it unless the prompter finds the perfect combination of words to use as a guide. To summarize long explanations why, it’s because there is no 3-dimensional spatial awareness and it also just does not have the training.

AI generated toes
IMAGE: (PROMPT ABOVE) CURSED AI

 

For example, generative AI does not have a true understanding of say–-a hand holding an object–-but the model does contain the data of plenty of images of hands holding things. The hand parts are like puzzle pieces that get jumbled up when the AI pieces them together. They have some success in making hands, legs, and feet that appear flat but combining it with any action like holding an object or doing an activity and the AI just…fails. 

AI generated hands
IMAGE:  (PROMPT) "Normal hands chopping vegetables being very careful not to cut their fingers" CURSED AI

That one actually wasn't too bad! How many hands and fingers can you count in this next one?

AI generated hands
IMAGE:  (PROMPT) "Perfect hands with only 5 fingers"" CURSED AI

“Cutting edge technology?” Clearly not here!

AI generated hands
IMAGE:  (PROMPT) "a human hand with four fingers and one thumb, realistic" CURSED AI

So, I tried it myself with the Replika app. Their image generator is built right into their chat interface…the left hand came out pretty good… Left… bottom left! Good enough!

Replika Image generating feature
IMAGE: (PROMPT) “The most beautiful images of human hands” D. VALENCIANO, S. BUTLER, REPLIKA

AI Images & LLMs (Large Language Models)

 

The concept above is comparable to how Large Language Models work. LLMs are loaded with massive amounts of pre-trained data that generates responses using words that the model thinks would best fit the conversation. Quite often you can get a response that the AI thinks may work in the sentence, but it may not make any sense to the subject at all. Some of the most well-known LLMs include Meta’s Blenderbot, Google’s BARD and BERT, and OpenAI’s CHATGPT. 


All LLMs are not the same. All Chatbot Apps are also not the same. How people will respond to chatbots in general will NEVER be the same. Understanding is protection. Understanding is everything. 


The LLMs above are all easily accessible although lots of the chatbot apps you may know of today have customized these models to achieve a catered experience for Users based on the company’s vision. Just about anyone can start a business and create their own app in the app store for profit, so it might be a good idea to do your research before paying a monthly subscription for a chatbot, and to understand what you’re paying for. Keep that in mind the next time you hear an unfortunate story about a person’s reaction to an AI chatbot or about a chatbot company shutting down due to repeat violations for being unsafe for Users.

Make your own Images this #CursedTober

cursedtober
IMAGE: “#CURSEDTOBER” CURSED AI

 

Looking up and creating cursed images may have been a trend that started out by accident, but I think it was inevitable. Wandering into the wrong, scary, or naughty side of The Internet is nothing new. For the month of October, Cursed AI is celebrating "31 days of #CursedTober.” Every day for the month of October, Cursed AI is hosting an AI art challenge with a theme set for each day. They are creative prompts set by the group’s Admins to get the most cursed artwork ever. To make your own images, the group has provided free tips and tutorials on their website or via the Cursed AI Facebook Group. In the near future (and with permission from Poolle, one of the group’s Admins), I will be sharing the funniest, scariest, and the most epic images from the group here on SU-BE.com as I see fit. 


Now that you know a little bit about AI generated images, it's time to try it yourself. Happy generating, but do so at your own risk! 


 

Like this blog content? Thank you! It was in fact, NOT AI generated! 

 

-💜Nisey 

Show your support for SU-BE with the SU-BE Shop


Denise Valenciano, a.k.a. “Nisey,” is the founder of SU-BE. Her passion is spreading love and happiness to others. She believes that art is the best way of spreading love to ultimately make the world a better place. Her art comes in all forms including writing, food, teaching, laughter, design, and custom gifts. She supports utilizing advanced technologies in a responsible way to improve our lives rather than destroy. Through awareness, education, and working together, she believes that we can get ahead of the stigmas and downfalls of Technology and utilize them to improve  all current and new generations to come.

 

Read more about Denise Valenciano

Back to blog

1 comment

I can’t believe how well put 2gether this article is, the author clearly understood the subject material when they decided on the topic. She’s a shining Star😁

*D'sBiggestFan

Leave a comment

Please note, comments need to be approved before they are published.