'Demonic' AI generating its own secret written language that nobody can understand

An image AI called DALL-E has sparked debate among AI experts who claim it is creating a secret language to categorise images - the words look like gibberish but have a hidden meaning

A popular AI tool that turns text into images appears to be creating its 'own language' of indecipherable gibberish and using it to categorise different pictures.

The DALL-E tool, which uses AI to generate images from text, is seemingly generating nonsense text when instructed to create images featuring printed words.

Computer science PhD student Giannis Daras took to Twitter to share examples of the 'language', including phrases the AI had created to identify birds and insects.

"Apoploe vesrreaitais" means 'birds', while "Contarra ccetnxniams luryca tanniounons" means insects.

Daras claimed in the viral thread that if you enter the gibberish words created by the AI back into the system, it will generate images linked to those phrases.

In a research paper which is yet to be peer-reviewed, Daras and his colleague Alexandros G. Dimakis said: "“[T]ext prompts such as: ‘An image of the word airplane’ often lead to generated images that depict gibberish text.

"We discover that this produced text is not random, but rather reveals a hidden vocabulary that the model seems to have developed internally.

“For example, when fed with this gibberish text, the model frequently produces airplanes.”

The academics believe that the AI generates its own words to make sense of the images it creates, and can then understand these words when they are read back to it.

https://twitter.com/giannis_daras/status/1531693093040230402?s=20&t=MTalqL01cueq575nKjzngg

The claims sparked debate on Twitter. One user, Dmitriy Mandel, said: "Hmm, time to brush up on signs of demonic possession."

Another person said :"Is this an actual LANGUAGE? With grammar and stuff?"

However, other AI experts remain highly sceptical of the claims.

Thomas Woodside said: "I do not believe this is accurate. At the very least, it is a lot more complicated than this thread / paper makes it out to be," before sharing a thread where he explained issues with the claims.

Another user went further, saying: "This is not science, this nonsense is **tarot card reading**... trying to find meaning in random noise."

DALL-E is a machine learning system that allows you to generate images just by typing short descriptions into a text box. You can find out more about it here