Here we are, with another year full of unnecessary artificial intelligence advancements. Apparently, no one in this field has seen any of The Terminator films at all. Despite our completely normal fear of a robot takeover, A.I. programs have now been able to create their own language when speaking to one another and projecting images of things they see in the world. This new language was discovered by scientists in California, as a new A.I. program called DALL-E 2 has been creating images based on text prompts. You can see this new language below:
From the above images provided by DALL-E 2, the artificial intelligence program has created a bunch of jumbled text to identify birds, and insects and then blend them together to see birds eating insects. While this might not sound threatening in any way, it’s also a program that is creating a way to identify real-life objects on its own. When tasked with showing “two farmers talking about vegetables, with subtitles” the program showed the image with a bunch of nonsensical text. However, it did identify the vegetables from a previous image that had been presented to the program.
Scientists are attempting to claim that it’s not a secret language, as in if this artificial intelligence program is going to be able to communicate with other programs. However, it is starting to develop its own vocabulary to correctly identify images that it had previously been shown. That might alleviate some of the concern, but if a program can identify threats via its vocabulary, things might get a little scary. Scientists have already created robots that can lift heavy items, jump high, not be knocked over, and identify people through a thick forest. Adding a “language” program to that might see these robots identify humans a lot quicker. We know what these scientists are up to.
What is more interesting about this artificial intelligence program creating its own language or vocabulary, is that the random gibberish text is not all that random. DALL-E 2 had been shown plenty of language data that didn’t just involve English, which made the images that the program identified with the text more accurate. Based on a word that the program produced, “Apoploe,” was used to create images of birds. Though this looks to be nonsense, the Latin word “Apodidae” refers to a genus of birds. So, this program was basically able to easily identify birds in some fashion. That is astounding when you think about it.
Though there are concerns that this artificial intelligence can be deemed “unsafe” scientists have assured everyone that DALL-E 2 is being used to test the practicality of learning systems. Apparently, if a program can be used to identify language parameters, then that learning system might be usable for children or those who are learning a new language, for instance. This “language” that the program has created is more about producing images from text instead of accurately identifying them every time. The program cannot say “no” or “I don’t know what you mean” so it produces an image based on the text it is given no matter what. We still think the robot uprising is going to happen.