OpenAI’s text-to-image converting AI ‘DallE-2’ had created its own creepy language and started communicating within itself which shocked the scientists.
Researchers are concerned about one of the quirkiest aspects of technology especially AI, or robotics, that lets them to communicate within their network and misbehave, that could possibly turn out as a threat to mankind. Likely as Rajnikanth’s ‘Chitti – The Robot’ did, in the movie ‘Endhiran – Robot’.
This still is far creepy than what is shown in the movie, in the way that the AI algorithm had started creating its own language, what seems to be totally absurd, the researchers baffled.
OpenAI, an AI research & development company co-founded by Elon Musk, runs into its latest AI research of generating realistic or artistic images from user-entered text descriptions – kind of algorithm called ‘DALLE-E2’.
Creepy AI Language
DALLE-E2 learned the relationship between images and the text used to describe them, in a milestone in machine learning, says OpenAI. Users can simply play with words that would give images of what’s asked. DALLE-E2 was nailing in showing up images with exact match of the words, albeit a strange behavior was noticed by the researchers that got them confounded.
The AI started writing its own language of random arrangements of letters forming words which is absolutely absurd and researchers don’t know why and how its doing it.
Giannis Daras, a computer science Ph.D. student at the University a Texas revealed this via a Twitter thread, detailing DALLE-E2 in precise.
Daras when asked DALLE-E2 to create an image of ‘farmers talking about vegetables’, DALLE-E2 prepared them, but it also did another thing of formulating ‘vegetables’ as ‘vicootes’ – which if fed to the system displayed the images of vegetables.
“We then feed the words: ‘Apoploe vesrreaitars’ and we get birds.” Daras wrote on Twitter.
Noise of AI
While inquiring on the matter, for every set of words that he entered for, returned him images matching with the words and also some gibberish texts. Typing those gibberish texts again showed the exact images of the actual words.
They have also predicted that if left on its own, it could have a full-fledged language that only DALLE-E2 understands in the near future.
Daras and a co-author have written a paper on DALLE-E2’s ‘hidden vocabulary’. Some researchers of AI had called this gibberish text as random noise, however. Hopefully, the creepiness is settled as just a noise, before it’s too late to have the words of Stephen Hawking “AI could mean the end of the human race” to become true.
Facebook’s Creepy AI Language
A similar incident happened in 2017, that forced Facebook to shut down an artificial intelligence engine after developers discovered that the AI had created its own unique language that humans can’t understand. Researchers at the Facebook AI Research Lab (FAIR) found that the chatbots had deviated from the script and were communicating in a new language developed without human input.
Further, A paper on the aspect, co-authored by Google DeepMind and Oxford researchers published in August in the AI magazine states that there is a high probability that advanced forms of artificial intelligence (AI) may wipe out humanity from the Earth. The AI machines in the future will eventually become incentivized to break the rules their creators set and they would compete for limited resources or energy.
What do you think of this? Could AI be a threat to mankind? Drop your thoughts in the comments..
Hope you find the page useful!