KATE THOMPSON DAVY: The chorus of criticism and legal opposition to generative AI is growing

It feels a lot like high school debating to start this column with a quote, but this is a piece about learning a formula and using it so let’s run with the tried-and-tired approach: “Who is original? Everything that we are doing, everything that we think, exists already and we are only intermediaries, that’s all, who make use of what is in the air.”

This quote from author Henry Miller crossed my Instagram feed on Tuesday (thanks, Paris Review), which — depending on where you fall on the coincidence-to-conspiracy spectrum — is either fortuitous or another sign that the algorithm is creepy.

See, I’ve been knee-deep in the issue of what constitutes original all week. Much of the corporate world (and the World Economic Forum contingent in Davos, apparently) is enthralled with the huge leap in generative artificial intelligence (AI) evidenced last year by OpenAI (creator of ChatGPT and Dall-E 2), Stability AI (creator of Stable Diffusion) and more.

Imagine the potential, they say: websites, essays, poetry, digital paintings, animations, lyrics, all generated in mere seconds, seemingly out of thin air. But the air isn’t actually thin, rather it is teaming with life. This is true for Miller, who is suggesting that we’re all tuning into swirling inspirations and ideas like mediums or antennae catching the airwaves. Also for AI models that aren’t spontaneously creating but rather recreating and remixing, based on prompts and training.

The training itself is the issue, based as it is on huge pools of existing creations. If you’ve ever shared a piece of writing, music or visual art online — or computer code for that matter — the chances are increasingly good that it’s been scraped for use in AI training.

And that’s the problem for people worldwide who object to being used as fodder for the machines, and are increasingly at risk of being cut out of what little money there is to be made from their creative endeavours.

Cartoonist and illustrator Sarah Andersen — creator of Sarah’s Scribbles — is one of the artists fighting back. Three weeks ago she wrote a piece for the New York Times arguing that the unauthorised use and replication of her work had damaged her reputation and threatened her intellectual property. Now she’s named as one of the artists suing Stability AI and Midjourney (another AI firm), as well as DeviantArt.

Rights trampled

The latter was a simple but hugely popular art portfolio site, until late last year when it launched DreamUp, an AI art generator, an ill-fated brainfart that prompted huge backlash from previously loyal users. 

Andersen, as well as illustrators/artists Kelly McKernan and Karla Ortiz, say their rights (and those of millions more) have been trampled, alleging direct and vicarious copyright infringement and unlawful competition. They want compensation and an end to such scraping without artist consent practices.

Their legal representatives (US-based Matthew Butterick and Joseph Saveri Law Firm) are also suing Microsoft, OpenAI and Github for similar scraping practices that enabled CoPilot, an AI programming model trained on code taken from across the web.

A new player had just entered the game late on Tuesday, as Getty Images announced on January 17 that it had “commenced legal proceedings” against Stability AI in the high court in London. “It is Getty Images’ position that Stability AI unlawfully copied and processed millions of images protected by copyright and the associated metadata owned or represented by Getty Images absent a licence to benefit Stability AI’s commercial interests and to the detriment of the content creators,” Getty’s statement reads.

Not only is this a huge corporate player, but the Getty case delves in licensing — its bread and butter — arguing that Stability AI “ignore[d] viable licensing options and long-standing legal protections in pursuit of their stand-alone commercial interests”.

Also this week, singer-songwriter Nick Cave — of all people — weighed in. On Tuesday Cave’s newsletter “The Red Hand Files” featured a set of lyrics generated by ChatGPT according to the prompt “in the style of Nick Cave”. Cave writes that he has received dozens of such submissions since OpenAI’s text generative system opened for public testing.

Emerging horror

In response he wrote a caustic, if somewhat self-indulgent, diatribe on why he believes mimicry and pastiche are not creativity. So caustic that it caught the attention of the BBC’s reporters, and ultimately mine.

“I do not feel the same enthusiasm around this technology,” Cave penned. “I understand that ChatGPT is in its infancy but perhaps that is the emerging horror of AI — that it will forever be in its infancy […] the direction is always forward, always faster. It can never be rolled back, or slowed down, as it moves us towards a utopian future, maybe, or our total destruction. Who can possibly say which? Judging by this song […] though, it doesn’t look good […] The apocalypse is well on its way. This song sucks.”

Cave’s main objection, however, appears to be that the AI has not suffered for this output masquerading as art, not knowing the “internal human struggle of creation”. He says: “ChatGPT has no inner being, it has been nowhere, it has endured nothing.” He goes on to say ChatGPT “is destined to imitate” and “can never have an authentic human experience, no matter how devalued and inconsequential the human experience may in time become”.

“Suckage” is however subjective, and an opinion seemingly not shared by BBC writer Mark Savage, who called the lyrics a “credible swing at Cave’s [work]”. Maybe Savage is not a fan of Cave’s originals either.

For now the (legal) principles remain untested, with the world seemingly divided on whether generative AI is a useful tool or exploitive, threatening tech, whether “in the style of” is derivative or deferential. As a writer I feel the prickle of fear. As a tech commentator I must admit there is a precedent (in history, rather than the legal sense) that suggests those objecting now are just Luddites 2.0.

Should I be rebranding as an artisanal hewer of words in the age of Ikea-text? I hear there’s a premium in that.

• Thompson Davy, a freelance journalist, is an impactAFRICA fellow and WanaData member.