Month: January 2023

Shutterstock, one of the internet’s biggest sources of stock photos and illustrations, is now offering its customers the option to generate their own AI images. In October, the company announced a partnership with OpenAI, the creator of the wildly popular and controversial DALL-E AI tool. Now, the results of that deal are in beta testing and available to all paying Shutterstock users.

The new platform is available in “every language the site offers,” and comes included with customers’ existing licensing packages, according to a press statement from the company. And, according to Gizmodo’s own test, every text prompt you feed Shutterstock’s machine results in four images, ostensibly tailored to your request. At the bottom of the page, the site also suggests “More AI-generated images from the Shutterstock library,” which offer unrelated glimpses into the void.

But, be warned before you jump on the chance to replace all your standard stock image favs with AI constructs: The idea of using artificial intelligence to pump out “art” is an increasingly divisive one. Generative AI is a landscape fraught with potential legal and ethical complications.

Why all the worry?

All AI is trained on datasets, i.e. massive aggregations of material that teach it what to aim for. And for AI image generators, those training sets contain images made by humans—often human artists for whom their work is their livelihood.

Multiple recent lawsuits have been levied against the AI art generator, Stable Diffusion, and others for copyright infringement. And there’s not yet a clear legal precedent for how these cases will be handled.

One of Shutterstock’s main competitors, Getty Images, has said it wouldn’t be wading into the murky waters of AI anytime soon. The site banned AI-generated images on its platform. And, with regards to the technology, Getty’s CEO, Craig Peters said, “I think that’s dangerous. I don’t think it’s responsible. I think it could be illegal,” in an interview with The Verge.

It’s obvious that AI must be pulling its “inspiration” from the work of real, live people. But it’s difficult to pin down exactly when and where AI generators steal from visual artists. Interpreting artistic style can seem subjective. On the other hand, AI’s acts of plagiarism are much more apparent—though no more egregious—in AI-produced text. Clearly, if not approached carefully, artificial intelligence could pave the way for a theft crisis in creative fields.

How is Shutterstock trying to get around the issue?

In an attempt to pre-empt concerns about copyright law and artistic ethics, Shutterstock has said it uses “datasets licensed from Shutterstock” to train its DALL-E and LG EXAONE-powered AI. The company also claims it will pay artists whose work is used in its AI-generation. Shutterstock plans to do so through a “Contributor Fund.”

That fund “will directly compensate Shutterstock contributors if their IP was used in the development of AI-generative models, like the OpenAI model, through licensing of data from Shutterstock’s library,” the company explains in an FAQ section on its website. “Shutterstock will continue to compensate contributors for the future licensing of AI-generated content through the Shutterstock AI content generation tool,” it further says.

The first pay-out to contributing creators was scheduled to be distributed in December, at the end of the company’s last fiscal quarter of 2022. It’s unclear how many contributors were paid last month, and how much was distributed, if any. Gizmodo reached out to Shutterstock with questions about this process, but did not immediately receive a response.

Further, Shutterstock includes a clever caveat in their use guidelines for AI images. “You must not use the generated image to infringe, misappropriate, or violate the intellectual property or other rights of any third party, to generate spam, false, misleading, deceptive, harmful, or violent imagery,” the company notes. And, though I am not a legal expert, it would seem this clause puts the onus on the customer to avoid ending up in trouble. If a generated image includes a recognizable bit of trademarked material, or spits out celebrity’s likeness—it’s on the user of Shutterstock’s tool to notice and avoid republishing the problem content.

But does it work?

As far as effectiveness goes, it took Gizmodo five different prompts similar to “robot drawing a picture of a robot” before the AI actually spit out something close enough to that concept. Again, each results page provides four AI-generated options. Of the twenty total images the machine generated, only the one included at the top of this post clearly showed some representation of a robot holding a drawing/painting. The others were… a mixed bag.

For now, and for the foreseeable future, I think I’ll be sticking to Shutterstock’s more standard offerings.

Early last year, the government of Bangladesh began weighing an offer from an unnamed Chinese company to build a smart city on the Bay of Bengal with infrastructure enhanced by artificial intelligence. Construction of the high-tech metropolis has yet to begin, but if it proceeds it may include face recognition software that can use public cameras to identify missing persons or track criminals in a crowd—capabilities already standard in many Chinese cities.

The project is among those that make China the world leader in exporting face recognition, according to a study by academics at Harvard and MIT published last week by the Brookings Institution, a prominent think tank. 

The report finds that Chinese companies lead the world in exporting face recognition, accounting for 201 export deals involving the technology, followed by US firms with 128 deals. China also has a lead in AI generally, with 250 out of a total of 1,636 export deals involving some form of AI to 136 importing countries. The second biggest exporter was the US, with 215 AI deals. 

The report argues that these exports may enable other governments to perform more surveillance, potentially harming citizens’ human rights. “The fact that China is exporting to these countries may kind of flip them to become more autocratic, when in fact they could become more democratic,” says Martin Beraja, an economist at MIT involved in the study whose work focuses on the relationship between new technologies like AI, government policies, and macroeconomics.

Face recognition technology has numerous practical applications, including unlocking smartphones, providing authentication in apps, and finding friends in social media posts. The MIT-Harvard researchers focused on deals involving so-called smart city technology, where face recognition is often deployed to enhance video surveillance. The research used information on global surveillance projects from the Carnegie Endowment for International Peace and data scraped from Chinese AI companies.

In recent years US lawmakers and presidents have expressed concern that China is gaining an edge over the US in AI technology. The report seems to offer hard evidence of one area where that shift has already occurred. 

“It bolsters the case for why we need to be setting parameters around this type of technology,” says Alexandra Seymour, an associate fellow at the Center for New American Security who studies the policy implications of AI.  

There is growing bipartisan interest in the US in restricting Chinese technology worldwide. Under president Trump, the US government imposed rules designed to restrict the use of Huawei’s 5G technology in the US and elsewhere and took aim at China’s AI firms with a chip embargo. The Biden administration levied a more sweeping chip blockade that prevents Chinese companies accessing cutting edge chips or semiconductor manufacturing technology, and has placed sanctions on Chinese providers of face recognition used to monitor Uyghur Muslims.

Further efforts to limit the export of face recognition from China could perhaps take the form of sanctions on countries that import the technology, Seymour says. But she adds that the US also needs to set an example to the rest of the world in terms of regulating the use of facial recognition. 

The fact that the US is the world’s second largest exporter of face recognition technology risks undermining the idea—promoted by the US government—that American technology naturally embodies values of freedom and democracy. 

When you tune into Twitch streamer Perrikaryal’s channel, you might see her playing FromSoftware’s role-playing game epic Elden Ring with fourteen, unfamiliar black sensors stuck to her scalp. It’s her—as she said during an informational stream earlier today—“just for fun” electroencephalogram (EEG) device, something researchers use to record the brain’s electrical activity, which she’s repurposed to let her play Elden Ring hands-free.

“Okay what and how,” publisher Bandai Namco responded to a clip of Perri (whose name seems to refer to the perikaryon, the cell body of a neuron) describing how she linked brain activity to key binds to help her play the game, shared by esports reporter Jake Lucky on Twitter.

Cue the disbelief (“I’ve gotten a lot of stuff online being like, […] ‘are you for real?’” Perri says in that Twitter clip) and cries of Ex Machina.

It does look incredible—in the clip, you see Perri simply say “attack” to her screen like a gamer girl Matilda and then, after a short delay, her Elden Ring character responds by casting Rock Sling at an irritated boss. But I spent my undergrad fixing eye-tracking devices to my friends’ heads while they helped me fill my lab requirements, and I know that, although brain technology can look complicated, some of it was still easy enough for me as a 19-year-old. So I reached out to my former classmate, University of Michigan cognitive neuroscience PhD candidate Cody Cao, for his thoughts.

“EEG has really good temporal resolution,” he said, “meaning that the collected neural response to gaming stimuli is down to milliseconds. If the neural responses corresponding to available actions present vastly different neural patterns, algorithms can decode or differentiate which is which after training. Then, you play the game with EEG.”

But playing a game with your brain—something Elon Musk tried to shock the public with in 2021, when his brain-computer interface company Neuralink released a video of a monkey playing Pong using its technology—won’t give you an advantage.

“Decoding is still janky,” Cao told me, “60 percent to 70 percent accuracy is considered pretty good,” compared to 90 to 100 percent accuracy in performing an action manually (which also requires your brain!).

“It takes algorithms a lot of training to get to an acceptable performance. They likely need to experience a lot of different examples of the same thing (like Perri saying ‘attack’ before attacking) to be able to account for a vast majority of attacks,” Cao continued. “It’s like FaceID on your iPhone—it gets better with the more examples it sees.”

Perri also emphasized in her stream today that she isn’t necessarily innovating, but bringing the possibilities of EEG usage to the general public’s attention.

“It’s not that crazy, it’s really easy to do. And it’s been done since 1988,” she said about gaming with her brain. “It’s not necessarily anything new that I’m doing, I’m just not sure that it’s very well known.” But now you know, and maybe you’ll figure out how to mind control me a grilled cheese that doesn’t hurt my stomach next.

Disrupting Earth’s chemical cycles brings trouble. But planet-warming carbon dioxide isn’t the only element whose cycle we’ve turned wonky—we’ve got a phosphorus problem too. And it’s a big one, because we depend on this element to grow the world’s crops. “I don’t know if it would be possible to have a full world without any mineral phosphorus fertilizer,” says Joséphine Demay, a PhD student at INRAE, France’s National Research Institute for Agriculture, Food and the Environment.

Since the 1800s, agriculturalists have known that elemental phosphorus is a crucial fertilizer. Nations quickly began mining caches of “phosphate rock,” minerals rich in the element. By the middle of the 20th century, companies had industrialized chemical processes to turn it into a form suitable for supercharging crops, hardening them against disease and making them able to support more people and livestock. That approach worked remarkably well: The post-World War II “Green Revolution” fed countless people thanks to fertilizers and pesticides. But sometimes there’s too much of a good thing.

We have liberated Earth’s caches of phosphorus so rapidly that the element now pollutes freshwater ecosystems, where excesses cause harmful algal blooms, infiltrates the snowpack, and decreases levels of dissolved oxygen in lakes and rivers. Studies suggest that humanity has grown too dependent on it for feeding the planet—and we are running out of this nonrenewable resource, which comes from geologic deposits that take millenia to form. When it washes from soil into waterways, it essentially disappears forever. A looming “peak phosphorus” moment threatens to increase prices and foment political tension if demand eclipses supply, as a large majority of reserves exist only in one corner of North Africa.

In a paper published this month in Nature Geoscience, Demay broke down how much phosphorus 176 countries have used between the years 1950 and 2017, and she estimated how much the use of mineral fertilizer contributes to soil fertility in each nation. Remarkably, phosphate rock accounts for around 50 percent of the world’s soil productivity. “It has never been quantified like that,” Demay says. And those numbers matter, she says, because “the work really highlights the high gap that exists between different world regions.” Wealthy countries in Western Europe, North America, and Asia use far more of the world’s phosphate rock than Africa, despite African soils being relatively deficient in it. “There is a need to distribute more equally the remaining first rock reserves,” Demay says.

James Elser, an ecologist with Arizona State University and the University of Montana who studies the global phosphorus cycle, was taken aback by that 50 percent figure. “That we’ve been able to mobilize phosphorus from these ancient geological deposits, and spread it around the world enough so that half of soil phosphorus is now comprised of industrial anthropogenic fertilizer, is pretty stunning,” he says.

And if the remaining supply goes down, prices will go up, exacerbating the access gap between rich and poor countries, says Dana Cordell, an associate professor and research director of food systems sustainability at the University of Technology Sydney. In 2008, phosphate prices spiked 800 percent due to supply and demand issues, and again 400 percent last year, due to Covid-related disruptions. The new study “shows how our global food system has now become heavily dependent on mined, nonrenewable phosphate rock,” she says. “And even if there is phosphate rock in the ground, it might not be economically viable to access it.”