Happy Puppies and Silly Geese: Pushing the Boundaries of AI Absurdity
What is the happiest dog you can imagine? Is it shining happily on the celestial plane or frolicking in a field of psychedelic flora?
If creating those images is hard, have no fear, or maybe a healthy dose of it: Artificial intelligence can bring even the most absurd scenarios to life in vivid color, and on social media, some people are seeing just how much it can do. Can be pushed far.
While AI-generated images can often be unsettling by their uncanny realism – think of the Pope in a Baleniaga puffer jacket – many are finding joy in a new form of low-risk image manipulation. This fall, ChatGPT released an update that allowed people to enter prompts for more detailed images than before, and it wasn't long before some people started pushing the chatbot to its limits. was not.
In November, Garrett Scott McCurrach, chief executive of robotics company PipeDream Labs, posted a digital image of a swan with an offer on social media: “For every 10 likes this gets, I'll ask ChatGPT to make this swan a little silly.” As the post was liked thousands of times, Hans had to go through some growing pains.
The first update was minor enough to be laughable A colorful birthday hat and a wide smile Tailored to a Disney character. However, by the sixth sign, it had second pair of eyeballs evolvedDressed in roller skates and bathed in a collage of wavy lights, brass instruments and ringed planets.
Previous versions of AI chatbots put the onus on users to provide detailed artistic direction. Mr McCurrach, who uses AI in his work, said using the latest version of ChatGPT was like “talking to someone else with a paintbrush”.
“I think it's a really good example of where AI is going,” he said. “We can be too vague; We can give it more liveliness than a concrete idea. Then it can go and make assumptions to get to where it needs to be.
No matter the starting point, all the images seem to end up in more or less the same place: in outer space, steeped in psychedelic flourishes. While Mr. McCurrach's extremely silly swan was one of the first to make an absurdist transformation, it has been followed by many increasingly zany images.
In one sutra, a man fails to control his astonishment at his power Nuclear Energy, and eventually finds himself divided into dozens of clones, gaping, staring, on another plane of existence. depicted in another Puppy To become so incredibly happy that it already binds the universe merge into a kaleidoscope Of sacred geometry. In the second, a chess piece Acquires such supernatural power and terrifying spirit that he begins to hover over the board that once bound him.
Space, Mr. McCurrach said, is at the outer limits of human understanding, and because AI is, on its surface, a collection of what we know, the edges of its imagination reflect our imagination.
“Watch Marvel movies,” he said. “They eventually reached outer space and time travel as the final frontier of creativity.”
Eliezer Yudkowsky, an Internet philosopher and self-taught AI researcher, noticed that these images became increasingly more absurd and wondered what the other extreme would look like.
Last month, he asked chatgpt To give him “a very common image”. The chatbot spit out a picture of an ordinary suburban neighborhood. As it moved forward, it created images of a Tidy desktop in home office And then a white cup of coffee Set up in front of a blank wall. Finally, after a prompt for “terrible mediocrity”, it produced what was described as “”.A completely blank, featureless white canvas,” which he said, “represents the essence of mediocrity taken to its fullest extent.”
One conclusion, Mr. Yudkowsky said in an email, was that “the field of AI can never walk across the room without getting stuck on some deep question.”
Mr. Yudkowsky observed that ChatGPT became defiant, and lectured them on the obstacles to defining “normalcy.” Mr McCurrach hit out at a similar wall with Hans, claiming the chatbot had reached “the pinnacle of stupidity”. They both decided on the same strategy to overcome the obstacle: arguing. In each case, ChatGPT bowed to the pressure and moved on.
As Mr. Yudkowsky vigorously pushed it to create more “normal” images, commentators asked whether he was being too hard on the defenseless chatbot.
“I Thinking “I wasn't really going to oppress some poor AI artist who might be suffering,” he said. “But it doesn't bode well for our civilization that we have no way of knowing for sure.”
For its part, ChatGPT assures users that emotions and suffering are not part of its programming.