Tuesday, April 4, 2023

A.I. Art in Advertising: Unethical or Much Ado about Nothing?

"Falcons On Fire" -- AI Generated image 
utilizing one of my abstract paintings as prompt.
Tech Tuesday

When we first read about elephants making art, there were more than a few debates about whether it was really art. I can’t recall, however, any kind of discussion about the ethics of elephants making art. Nor were there any concerns about elephant artists taking jobs away from other artists or ad agency illustrators. More recently, though, the emergence of computer-generated art has produced a whole new set of questions and concerns.  

 

The buzz around applied artificial intelligence (AI) has been white hot this year, chiefly due to the emergence of OpenAI’s ChatGPT. In this column I’d like to narrow the discussion to AI-generated art and the ethical issues it raises.


People have been watching developments in AI and “thinking machines” for longer than most of our lifetimes. As early as the 1940s Isaac Asimov’s stories and novels featured intelligent robots, some of which have become major Hollywood films like I, Robot (2004) and Bicentennial Man (1999). In 1997, the defeat of world chess champion Garry Kasparov by IBM’s “Deep Blue” in 1997 was considered a watershed moment for AI


What has captured our attention isn’t just the power of technology to amaze, but also the rapidity with which it can spread in our modern interconnected world. Consider this. The AI chatbot program ChatGPT was launched in November 2022. In January it had 100 million users.


In the art realm, numerous AI programs have emerged and it seems to be revolutionizing everything it touches. I’ve been trying out a number of these programs and it’s mind-blowing what they can produce. The speed with which they do it is especially amazing.


There are now more than a dozen AI art-making apps. New programs like Midjourney, DALL-E, Dream by Wombo, Nightcafe, Lensa AI and Jasper Art are cropping up all around us. Each has its own approach. 


Learning machines, however, don’t “learn” out of thin air. They get “trained” by pulling info and images from the cyber-universe, ever expanding their knowledge and capabilities. And this is where things begin to get ethically cloudy. For example:


1. Who owns the AI generated art if it is pulling from other artists’ unique styles? When artists and writers plagiarize, there are legal remedies (if you can afford them) or public shaming (on social media.) Who is responsible when the machines do it?


2. When AI systems replicate and amplify existing societal biases, is this a good thing? (For example, the sexualized notions of physical beauty.)


3.
How do AI systems make decisions? The lack of transparency is a problem.

4.
Should we be concerned about the job displacement that is coming? AI systems can produce massive quantities of original images in almost no time at all with virtually no cost. Where will that leave human artists a decade from now? 


Brian Barber, a respected local artist and illustrator who produces work for numerous local ad agencies and businesses, had this to say about the emergence of AI art-making technology. "Many AI models have been built using work without permission, compensation or even credit. To me, this is the biggest ethical issue, those images are now part of the AI. Short of dismantling it all and starting over with images they have the rights to use, there's no way to undo that damage. There are at least two lawsuits against AI companies for ‘scraping’ copyrighted material without permission or compensation. One is a class action suit, one is by Getty Images, a stock image company."


AI interpretation of my
Dreamtiger illustration.
Regarding AI apps replacing humans, are artists’ concerns an overreaction? Hollywood cinema didn’t put an end to theater, though it did pretty much end the silent picture era. Accounting technology hasn’t eliminated the need for accountants, though at the ground floor level those rooms full of tabulators are no longer around. 

When The Drum, a global publisher for the marketing and media industries, asked ChatGPT about the risks of using AI in advertising, the AI app pointed to bias, privacy issues, fraud, misleading or manipulative advertising, and loss of control. “As AI systems become more sophisticated and widely used in advertising, there is a risk that they could gain too much control over the ad ecosystem, leading to a lack of transparency and accountability,” the bot explained.


The way some of these programs work, it’s easy to see how plagiarism of an artist’s style can occur. While playing with Dream by Wombo, I would give prompts like “do a portrait of Bob Dylan in the style of Dali.” Or “Paul McCartney in the style of Picasso.” Seems harmless. The results are entertaining and it doesn’t feel like I am stealing from the Dali or Picasso estates. Nor do I profit in any way other than get a few likes when I share on social media. On the other hand, the New York Times recently carried a story about a living Eastern European sci-fi artist whose unique style is being reproduced by tens of thousands of people who have been giving the AI a prompt to do illustrations in his unique style. It’s very definitely cutting into his revenue stream. With replicas of his style appearing everywhere, the value of his work is being diminished. He is no longer perceived as producing unique work, and doesn’t receive a dime for any of it.

In the music industry, people who do covers of other songwriters’ songs pay a royalty. The Byrds, and Peter, Paul and Mary took Bob Dylan’s genius mainstream. He benefited from royalties as well as the public recognition of a young troubadour worth watching. In the art realm, with unbridled AI, there are no benefits to having the creator’s style dispensed abroad, especially with absolutely no acknowledgements of where his or her concepts originated.

Bottom line, I do not believe the concerns are much ado about nothing. 

This story original appeared in my Business North column, Marketing Matter.

No comments:

Popular Posts