Skip to main content

Like an A.I. acid trip, this neural net rebuilds reality with flowers and fire

gloomy sunday

When computers get creative, the results are frequently fascinating — as a new project created by artist and machine learning Ph.D. student Memo Akten admirably demonstrates. A bit like projects such as Google’s Deep Dream image generator, Akten has been applying artificial neural networks to create some unusual visual effects. His “Learning To See” project uses image recognition neural nets to interpret the images it sees on a live video feed. The twist? He trained his different neural networks exclusively on a diet of only water, sky, flowers or fire still images so that regardless of what image they actually see, they interpret it as waves crashing, fires roaring, or flowers growing.

“In some ways, this was a response to the binary polarization that we see politically in the U.K., in the United States, and in Turkey, which is where I’m from,” Akten told Digital Trends. “The idea is that all of us are only capable of seeing the world through the lens of what we’ve seen before. We incapable of seeing it through other people’s eyes because we’re so colored by what we know. In the case of this piece of work, the neural network has been trained only on certain images — such as waves or fire or flowers. As a result, everything it sees it can only make sense of based on its own experience.”

Related Videos

It’s an intriguing concept, both conceptually and technologically. Particularly impressive from a tech point of view is how fluid the movements look, despite the fact that Akten says the neural networks were trained exclusively on still images. Nonetheless, through analyzing only still images the A.I. has approximated a fairly accurate idea of how fires burn or water moves.

“With any emerging technology, artists will always think about how they can apply it to their own domain, whether that’s painting, dance, performance, or whatever else,” Akten continued. “Right now these machine learning technologies are still a bit complex and inaccessible for a lot of people. But there’s a lot of work being done to make these tools into things which can be used by everyone.”

Editors' Recommendations

Voice actors seeing an increasing threat from AI
best podcast apps for android microphone in radio station broadcasting studio

Voice actors are the latest creatives to feel the heat from artificial intelligence (AI).

According to a Motherboard report this week, those providing their voices for content such as ads, gaming titles, and animations have been noticing how clients are increasingly asking them to sign contracts that hand over the rights to their voice so that AI can be used to create a synthetic version.

Read more
Chinese internet giant to launch its own version of ChatGPT, report says
brain with computer text scrolling artificial intelligence

Chinese internet giant Baidu is planning to launch its own version of ChatGPT, a report claimed on Sunday.

The company will unveil its AI-powered chatbot in March, a person claiming to have knowledge of the matter told Bloomberg.

Read more
Here’s how ChatGPT could solve its major plagiarism problem
Close up of ChatGPT and OpenAI logo.

ChatGPT is a wonderful tool but there's a dark side to this advanced AI service that can write like an expert on almost any topic -- plagiarism. When students that are supposed to be demonstrating their knowledge and understanding of a topic cheat by secretly using ChatGPT, it invalidates testing and grading. AI skills are great but aren't the only subject that students should learn.

Policing this problem has proven to be difficult. Since ChatGPT has been trained on a vast dataset of human writing, it's nearly impossible for an instructor to identify whether an essay was created by a student or a machine. Several tools have been created that attempt to recognize AI-generated writing, but the accuracy was too low to be useful.

Read more