Ghost in the Machine

With AI-generated art on the rise, what does it really mean for humanity?

- Koyyi Chi

On September 2, one Jason M Allen of Pueblo West, Colorado, won the blue ribbon in the Colorado State Fair's contest for emerging digital artists.

While the public's reaction was understandably incendiary (which we will further expand on later), the state fair had a division for "digital art/ digital manipulated photography". Let's talk about why Allen hadn't technically broken the rules.

If we look at it from a simplified technical perspective of how the AI program works: after being prompted by a "command", the AI sets about collecting data from the Internet (in the case of an AI art generator, it's any kind of visual media you can find online); the more specific the command, the more refined the AI gets as it pieces together the command, therefore the closer the generated image is to the command. Instead of a slightly random version of "a child in a yellow raincoat holding a red balloon", you can type "a realistic oil painting in the style of Vincent Van Gogh of a child in a yellow raincoat and blue Wellington boots playing in the rain, holding a red balloon in their hand".

To further demonstrate, herein are some examples we've made of the prompts above with the free version of OpenAI's Dall-E mini, now known as Craiyon, in both the simplified version and the complex version:

Yes, we know, the melting faces in the portraits are haunting and the proportions are dubious at best-given that Craiyon is a simpler AI programme compared to considerable powerhouses like Dall-E 2 and Mid journey.

The point we're making is that because the AI is a simulation of how human intelligence works to a certain extent, these Al art generators are a mimicry as to how humans create artwork. Thus, making it a mere tool.