AI’s creative conundrum: From copyright infringement to artistic originality

Concerns over copyright infringement

Foong Cheng Leong, a lawyer in Malaysia specialising in intellectual property, explains that copyright owners have control over their content under copyright law. “AI companies that train their models on content without permission of the copyright owner, whether authors, artists or others, are infringing copyright. That is unless the gathered data is deemed fair dealing, such as for non-profit research depending on the laws of the relevant countries,” he says.

There are privacy issues to consider, too, he says and cites the example of celebrities being increasingly concerned about their image and likeness being captured by AI companies that can later be used to generate their performances with the click of a button.

In late 2023, various groups of authors, including Game of Thrones author George R R Martin as well as visual artists, music publishers and media company Getty Images, filed lawsuits against OpenAI, Stability AI and Midjourney for allegedly misusing their work to train genAI systems.

In Foong’s opinion, AI companies should acknowledge when their models utilise copyrighted or creative works. He adds that platforms could implement rules requiring users to agree to have their content used to train AI models, with the understanding that the original creators would receive some payment for their work. Establishing a system where AI companies must pay a fee to use creative works would help ensure that the creators are fairly compensated for their contributions.

Read more: Making the pivot: Why this health tech startup went into AI medical diagnosis

The difference is in artistic originality

Some artists like Abdul Hafiz Abdul Rahman appreciate the ability of AI to generate initial ideas and concepts for them. But the overreliance on AI, says the Malaysian creative who also goes by his moniker Katun, could stifle an artist’s creativity. “You will spoil yourself and won’t want to think anymore,” he says. “As long as you recognise that AI is a tool that helps you stay creative, I don’t see any harm in using it.”

He thinks the creativity an artist pours into the piece will set their work apart from those generated by AI. “When I paint, I implant things from my memory into my art, which makes it personal,” he says, adding that it imbues his art with a “soul” and makes it feel “alive”.

On the other hand, he says that because the works that genAI [tools] produce are just copies and recompilations of other artists’ work, “there is no originality at all”.

Chia compares AI to a savant artist, who possesses an impressive memory and is capable of recalling 100,000 images with ease, but its creative prowess faces limitations.

“The limitation of AI-generated art lies in its inability to venture beyond the known,” he says. “It can faithfully replicate what already exists but falters when asked to conjure the uncharted. Unlike a visionary artist anticipating the next breakthrough, AI remains tethered to its training data. It can mimic the artist’s style but lacks the foresight to birth entirely new masterpieces based on predictive leaps from the artist’s body of work.”

NOW READ

Why do Elon Musk and other tech leaders want to pause AI experiments?

AI doesn’t like to credit its sources. For artists, that’s a problem

Man vs machine: Is AI pushing creative boundaries or killing it?