First, A.I. came for the artists—here’s why that was a mistake

Of course, A.I. lacks consciousness and therefore intentionality. But if it were trying to act strategically, it would have made some serious blunders. Not long ago, there was broad consensus that creative work was relatively safe from automation, but—as we have discovered since OpenAI’s DALL-E 2 started generating at least superficially impressive imagery last year—that’s not the case. 

As it turned out, A.I. came first for the creatives—and boy do we know how to speak out. 

The thing is, generative A.I. is inherently plagiaristic, even if it often does a good job of covering its tracks. As my colleague Prarthana Prakash just reported, the popular science fiction and fantasy magazine Clarkesworld has suspended all new short-story submissions, because its inbox was inundated with A.I.-generated content that aped previously published material, with minor changes.

“The number of spam submissions resulting in bans has hit 38% this month,” wrote publisher Neil Clarke in a blog post. “While rejecting and banning these submissions has been simple, it’s growing at a rate that will necessitate changes. To make matters worse, the technology is only going to get better, so detection will become more challenging.”

See also: Dow Jones and CNN complaining about OpenAI training ChatGPT—its A.I. chatbot that can author reports—on their writers’ articles; Getty Images suing startup Stability AI because its popular Stable Diffusion image-generation tool allegedly scraped Getty’s pictures; and artists expressing outrage at users of generative A.I. using such tools to copy their styles to create art.

I’ve been thinking about these ethical issues a lot recently, not just because I’m a journalist—whose peers have recently been caught publishing error-strewn, A.I.-generated stories—but also because I’m in a band. 

Last year, we hit the stage for the first time since before the pandemic, and we needed to create a flyer to promote the event. Now, we’re just a hobby band and, being reasonably tech-savvy, we thought we’d make a flyer ourselves rather than paying someone to design the thing. So I fed some of our lyrics into Midjourney, an A.I. that can create art based on words entered into it, and got an image that I thought looked pretty cool; here it is. Months later, I saw some other band’s A.I.-generated thumbnail image that had an extremely similar figure in the foreground—and realized that both those images had to have been ripped off the same source.

So. Many. Ethical. Questions. Should we have paid someone to make real art for the flyer? We were unlikely to do so, but there are a bunch of graphic designers out there who are just starting out and have probably lost a substantial potential client base in the past year. Is it bad that our flyer ripped off someone else’s art, even unwittingly? Of course, and I’d apologize if I knew whose art that was. Might music-oriented generative A.I. one day rip off my creations? It’s quite possible, and I would not be happy about it.

As a musician, should I, therefore, draw a red line now and say I’ll never again use A.I.? I’m tempted to, though I also wonder if such resistance will one day look as futile as the rock band Queen promising to never use synthesizers, only to one day give in.

A.I.’s early “targeting” of the creative classes matters because, without meaning to sound pompous, artists and journalists have a loud voice and therefore play a significant part in shaping our collective consciousness. We’re not shouting about the latest developments in A.I. just because they’re fascinating, but also because we know deep down that our professions will soon be fundamentally changed by them. Yours is next, but at least society’s early warning system is sounding.

Want to send thoughts or suggestions to Data Sheet? Drop me a line here.

David Meyer

Data Sheet’s daily news section was written and curated by Andrea Guzman. 

NEWSWORTHY

Restrictions on Bing’s A.I. chatbot are loosened. On Friday, Microsoft limited users to asking Bing five questions during each chat session and only 50 sessions per day. Yesterday, the company raised the limit slightly to six questions per session with a limit of 60 sessions a day. According to a company blog post, the plan is to increase the daily cap to 100 total chats soon. But after reports of Bing engaging in strange conversations, Microsoft says it’s still working on how to manage the chatbot responsibly. 

FTC declines to challenge Amazon’s One Medical deal. The $3.9 billion deal for primary care provider One Medical has been under review by the Federal Trade Commission for eight months. But the deadline for suing to block the deal before the acquisition closes ended yesterday, Bloomberg reports. Still, the FTC says it plans to continue probing the deal after it’s completed. The acquisition is Amazon’s big push into health care, giving it more than 200 medical offices in 26 U.S. metro areas. 

Twitter employees are still getting cut. Dozens of employees across sales and engineering departments were laid off last week, reports The Verge. It’s the third round of layoffs since Elon Musk announced in late November that layoffs were over. 

Workers challenge return-to-office mandates. More than 14,000 Amazon employees have joined a “Remote Advocacy” Slack channel after CEO Andy Jassy called for a return to the office last week. Over at Google parent Alphabet, CEO Sundar Pichai was sent a letter yesterday by Sen. Bernie Sanders (I-Vt.) and Rep. Greg Casar (D-Texas) throwing their support behind YouTube Music workers who have been called back to the office in Austin. The RTO mandate came shortly after 58 workers filed a petition with the National Labor Relations Board seeking to join the Alphabet Workers Union. The two lawmakers wrote that they were worried the in-person work requirement came in retaliation to the workers’ organizing efforts.

FOOD FOR THOUGHT

WFH helps some people boost their savings. In New York City, workers are spending almost $5,000 less per year on expenses near their offices. Reduced spending has also reached other metro areas where remote work is common owing to white-collar workers staying home and extended pandemic restrictions. A new report from Work From Home Research shows that people in big cities have been able to pocket thousands of dollars a year by not going into the office. 

From the article: 

WFH Research’s calculations only consider spending near the office. It doesn’t even factor in other expenses saved when working remotely, such as buying fancy clothes or cosmetics, Nick Bloom, professor of economics at Stanford University and one of the researchers behind the report, told Fortune. Working from home can also save time and offer convenience, he added, such as being around to let in domestic workers (like a plumber) and getting back time spent commuting.

IN CASE YOU MISSED IT

Coinbase beats revenue estimates but still posts $2.6 billion loss for 2022 amid Crypto Winter, by Leo Schwartz

Bernie Sanders sides with Bill Gates and says he wants to tax the robot that takes your job, by Tristan Bove

Google Cloud becomes Tezos validator in latest push into Web3, by Ben Weiss

Biden’s push to make America a leader in EV manufacturing may finally be paying off as Toyota announces a plant in the U.S., by Prarthana Prakash

Elon Musk wants customers to trial SpaceX’s Global Roaming Starlink service. The only catch is it will cost you $3,000 for the year, by Eleanor Pringle

BEFORE YOU GO

YouTube Music’s newest feature. Users can now create their own radio stations by picking up to 30 artists they want to hear, adjusting the frequency of these artists in the mix, and applying filters so that the station emphasizes playing new releases. The feature is different from what’s available on Spotify and Apple Music. For now, radio features on those platforms create a mix after a user chooses a specific song or musician.