Artificial intelligence writes fairy tales for children, gets interrupted on Twitter: “Too bad”

Reshi decided to write a story using ChatGPT and Midjourney’s AI. After posting her result, she was flooded with hateful comments.

Ammaar Reshi was reading a bedtime story to her friend’s daughter, and 72 hours later she was able to publish her AI-written and illustrated book for free.

Impressed with the results, Reshi shared the experience on Twitter, which garnered over 2,000 comments and 5,800 retweets. At first, positive reviews appear, then a real war begins under Reshi’s writing.

The story “Alice and Sparkle”.

It all starts with a children’s tale that tells the story of Alice and her friend Robot Spark, using the new OpenAI chatbot, ChatGPT and Midjourney to write and describe Resh. He is not alone, even Roy Ming team in Italy experimented with artificial intelligence to write fairy tales. Many play with new technologies. Satisfied with the result, Reshi then decides to publish it on Amazon, give it to his friends’ children, publish the success on social networks, and then something goes wrong.

“I woke up at 4am with my phone exploding, with a new tweet every two minutes saying things like ‘You’re evil’ and ‘We hate you,'” Reshi explained. His other was just an experiment to share with friends, but it was the starting point to start a rebellion on Twitter.

The problem with the new AI

Reshi’s book created a huge buzz. Increasingly powerful artificial intelligence is perceived as a threat. Some see the beginning of the end of all white-collar jobs and see the chatbot as a harbinger of mass unemployment. However, artists were the first to point the finger at tools like Midjourney or ChatGPT.

For many, these are a real steal. To work, the AI ​​must be trained with large amounts of data, including artists’ work downloaded from the Internet and uploaded to Dall-E, Midjourney, or ChatGPT without permission. Users can also enter the artist’s name in the query to illustrate, for example, “in his style”.

For example, Lensa AI, which creates “Magical Avatars”, i.e. user selfies that turn into works of art, was at the center of the discussion. Australian artist Kim Leutwyler accused the program’s developers of allowing the algorithm to use and reproduce their own artistic styles: “I was immediately suspicious when I started seeing all these portraits created by some of my friends and even other artists using Lensa. Leutwyler told the Guardian. “Some of the works are clearly related to the work of other artists. »

Reshi’s reaction

“I hadn’t read about the issues,” Reshi said. “I realized that Lensa was actually causing all of this by being a very widespread program. He spread this debate and I was just getting hateful comments about him. »

“I was just shocked and honestly I didn’t know how to handle the situation. Among the negative messages, Reshi explained, she also reads reasonable comments that explain exactly what the real problem is. “These are the people I want to communicate with,” he said.

Second tweet

After reading and listening to the issues listed below, Reshi added in a tweet: “I think artists should be involved in creating AI image generators and their talent, skills and hard work should be respected.”

He also explained that all the hateful comments were misdirected and that it was his one-off project. Reshi explained that he never actually claimed to be the author of the book. “I wouldn’t call myself an author at all,” he said. “The AI ​​is basically a ghostwriter and the other AI is an illustrator. Spent hours tweaking Midjourney’s suggestions to try to get consistent artwork.

Amazon review

Amazon suspended sales of Reshi’s book from January 6 to 14 due to “suspicious research activity” attributed to review volumes ranging from one to five stars.

Some people have indeed criticized the quality of the book’s writing and illustrations. “The writing is harsh and the sound is lacking,” read one of Amazon’s comments. “And the art—wow—is so bad it hurts. Tangents everywhere, weird fingers on every page, and inconsistencies that make these pictures feel like they’re a cut above chance. »

Half the experience

Reshi left the experience frustrated, explaining that he would not be releasing a new picture book, but that he was still interested in testing the AI. “For example, I would use ChatGPT,” he explained, “currently there are fewer content ownership concerns than AI image generators. He added that the goal of the book was always to give it to his friends’ children, “we both liked it, with people I mean it worked and was fantastic.”

Leave a Reply

Your email address will not be published. Required fields are marked *