Artificial Intelligence Caught Writing Its Own Creepy Language By Researchers
Nobody else is capable of fully understanding the language coined by OpenAI's "DALLE-E2" artificial intelligence system
Something creepy recently happened in the world of technology after an artificial intelligence programme reached the pinnacle of independence by writing its own language.
Nobody else is capable of fully understanding the language coined by OpenAI's "DALLE-E2" artificial intelligence system. Its job is to generate realistic and/or artistic images based on text descriptions entered by users. OpenAI claims that DALLE-E2 is groundbreaking, for it effectively "learned the relationship between images and the text used to describe them."
iStock
DALLE-E2's secret language
While all this sounds riveting, DALLE-E2 is on a secret mission. It is currently writing its own language - a random string of letters. The worst part is that researchers don't know why something like this is happening.
The finding was made public by Giannis Daras, a computer science PhD student at the University of Texas. Daras had directed DALLE-E2 to set up an image of "farmers talking about vegetables." The programme did that, but it sneaked in a word called "vicootes" - which researchers say belongs to its self-written language.
Also read: Hate Waiting? AI Can Eliminate Waiting Time At Red Signals And Declutter Roads
When Daras checked the programme for "vicootes," they saw pictures of vegetables. That's not all! Wonder what birds are called in this AI language? "Apoploe vesrreaitars," according to Daras.
DALLE-2 has a secret language.
¡ª Giannis Daras (@giannis_daras) May 31, 2022
"Apoploe vesrreaitais" means birds.
"Contarra ccetnxniams luryca tanniounons" means bugs or pests.
The prompt: "Apoploe vesrreaitais eating Contarra ccetnxniams luryca tanniounons" gives images of birds eating bugs.
A thread (1/n)? pic.twitter.com/VzWfsCFnZo
Daras, along with a co-author, has now written a paper on DALLE-E2's "hidden vocabulary." It seems that the AI is attempting to create a world of its own as it performs its job. Each time you throw a command its way, it might respond with some gibberish text. When researchers ran this gibberish on DALLE-E2, they found images that signal what the word means.
iStock
A few people, including @realmeatyhuman, asked whether our method works beyond natural images (of birds, etc).
¡ª Giannis Daras (@giannis_daras) June 3, 2022
Yes, we found some examples that seem statistically significant.
E.g. "doitcdces" seems related (~4/10 images) to students (or learning). (3/N) pic.twitter.com/gtfywmRBkD
Also read: Worried About Being Spied On By Your Tech? New AI Algorithm Can Stop Eavesdropping
iStock
If left to its own devices, we could have a full-fledged language that only DALLE-E2 understands in the near-future.
What do you think about AI writing its own language? Let us know in the comments below. For more in the world of technology and science, keep reading Indiatimes.com.
References
Baum, T. (2022, June 3). Artificial intelligence spotted inventing its own creepy language and it¡¯s baffling researchers. . . The Sun.
I live for all-things-technology - gadgets, novel climate solutions, and startups that are changing the game. In my leisurely hours, you can find me binge watching science fiction films, writing poetry, or dancing to pop anthems.
Read More
