AI Art Is Now a Thing

Intrigued by AI-generated art? There’s now an online marketplace for users to purchase text descriptions for AI models such as DALL-E 2 to generate highly realistic, original images.

Known as prompts, the idea is to curate detailed instructions that are interpreted by DALL-E 2 to reliably generate specific styles or categories of images. Launched in June, PromptBase allows a “prompt engineer” to put up their prompts for sale.

While PromptBase only hosted prompts for DALL-E 2 and GPT-3 when it was first launched, it now supports Midjourney and Stable Diffusion. It is understood that PromptBase takes a 20 percent cut of sales.

So, is a prompt engineer an artistic or engineering endeavor? The Verge posted an interview with prompt engineer Justin Reckling last week, which shed some light on the matter. Somewhat unsurprisingly, Reckling, who got into prompt engineering in April this year, has a background in programming and software quality assurance.

“I see prompt writing from the perspective of an artist, coder, and engineer. I use my programming experience to help me understand how the service may interpret my prompt, which guides me to more effective tinkering with it to coax the results I’m after,” said Reckling.

“Every word in a prompt has a weight associated with it, so trying to work out what works best and where becomes a core asset in the skillset. My background in software quality assurance is a pretty big driver in that ‘what happens if’ style of thinking,” he said.

Part of the research Reckling embarks on sees him asking subject matter questions to GPT-3 to glean additional details. To create a prompt about someone getting a manicure, for instance, he might ask “Someone is getting a manicure done; explain what you’re seeing.”

Reckling appears to be doing well for what is presumably a side hustle. With an inventory of 50 prompts, he sells three and five prompts every day. It is not without cost, however. Reckling says he spends up to USD15 in credits for each prompt he creates. For now, the Block Cities prompt (below) has received the most sales.

AI-generated artwork wins contest

In the meantime, an AI-generated artwork has won first place at the Colorado State Fair’s fine art competition.

Jason Allen, a game designer, won first place in the emerging artist division for digital arts or digitally manipulated photography category at the Colorado State Fair Fine Arts Competition.

His winning entry, titled “Théâtre D'opéra Spatial” or “Space Opera Theater” was generated using Midjourney – which beyond netting him a USD300 prize, also generated a fair amount of controversy.

“This sucks for the exact same reason we don't let robots participate in the Olympics,” one Twitter user lamented. On the other hand, the industrial revolution would never have happened if we stuck to making everything by hand.

Though Allen didn’t use a paintbrush, digital or physical, to create his submissions, he apparently spent more than 80 hours working on some 9,000 iterations of prompts to refine the final three submissions – including the winning entry.

“Rather than hating on the technology or the people behind it, we need to recognize that it's a powerful tool and use it for good so we can all move forward rather than sulking about it,” Allen told CNN.

At least one of the judges that CNN spoke to did not realize the piece was generated by AI when judging, though he maintained his decision when told about it.

The debate over AI-generated art won’t be resolved anytime soon, but what is clear is how AI-generated art is now a thing.

What caught my attention though, was how Allen improved the mediocre resolution of his Midjourney-generated image for his submission. Yes, it was upsized by another AI-powered software known as Gigapixel AI.

Paul Mah is the editor of DSAITrends. A former system administrator, programmer, and IT lecturer, he enjoys writing both code and prose. You can reach him at [email protected].

Image credit: Screenshot/Discord