Artificial intelligence (AI) is widely used in polar scientific research and is proving to be a valuable tool, particularly in data collection and analysis. But there is a dark side to this shining medal. AI-generated videos are flourishing on the internet, promoting an inaccurate and stereotypical image of the polar worlds that risks replacing the discourse of polar experts.
A group of bearded men in parka suits, polar bears almost smiling, and music from the film Titanic in the background. There’s no doubt about it: this is a video generated by artificial intelligence. In these utterly improbable, repetitive sketches, we see the King of the Arctic behaving like a big doggo. And he’s not the only one involved. Penguins, walruses, arctic foxes, belugas and other whales also get their share of footage. Between rescues, cuddles and utopian interspecies encounters, AI is invading the polar worlds, not without consequences, especially for the professionals who work in these regions. Behind the scenes, a kind of reconciliation with nature could explain the success of these videos among a public new to polar issues.
How to explain the virality of these videos, which are not always very well made, to this audience? “There is an emotional dimension to these images”, explains Olivier Glassey, a sociologist specializing in digital uses at the University of Lausanne. For him, these videos show a form of reconciliation between man and nature that is bound to strike a chord. “We know that what works well on social networks, the main vehicle for virality, is emotion.”
Added to this is the “cute” aspect, very popular on social medias. Animals are depicted as cuddly toys that humans hug as if they were domesticated but for the beauty of the figure. “It is the instrumentalization of an emotional form that focuses on something positive in order to create traffic, views or redirection of information”.
Scientists and photographers put to shame by AI?
However, these images are not without consequences for professionals working in the polar regions. The issue is the ability of artificial intelligence to generate an immense amount of content on a daily basis, which could, for example, drown out the work of photographers. “In the comments, we see reactions from photographers specializing in the polar regions, who are very angry that people are being fooled when, for them, it’s often the investment of a lifetime to produce these images. The AI images, with the ease with which they can be produced and despite their flaws, are sometimes more viral than their photographs”. Diluted in a mass of artificial images, professional shots could well lose much of their audience, but also their relevance. “It’s something that has to be quite violent symbolically to be experienced. The people I saw reacted strongly to it, and understandably so. These images replace the possibility of bearing witness to reality.
The same problem applies to scientists, who will probably have to coexist with this artificial content on social networks, at the risk of having the public’s attention diverted from their research. “Social networks are a battleground for people’s attention. All this artificial content can saturate and overwhelm the virtues that some of these vectors could have for simply finding out what’s going on, discovering new things or feeding a scientific curiosity,” notes the sociologist. For example, a video produced by a scientific expedition could be drowned out by fifteen machine-generated videos. While the public might think that these videos are all about the same thing, they might also prefer the AI content… because the colours are warmer and brighter. So it’s up to scientists to rise to the challenge of developing strategies to attract attention without losing their scientific integrity or the nuance and complexity of the information they provide.
Another problem for polar scientists and experts, especially wildlife conservationists, is the misrepresentation of their work. In the AI videos, we see wild and dangerous animals being handled without precaution. There’s no sign of anaesthesia and not a gun in sight in these videos, which show a docile and very approachable nature. “The most striking thing is the idea of a wilderness in complete harmony with man. All the difficulties of working in the field and what it means to approach this wild world while respecting it are completely erased.”
The power of the mass
But why so many videos? Is it a way to train AI generators by producing mass polar bear rescues? “There are quite a few Internet players playing with the new possibilities of artificial intelligence. Hundreds of thousands, or probably millions, of people are experimenting with the possibilities that have emerged in the last few years,” notes Olivier Glassey. “These are people who are trying to find ways to automate a number of processes, such as generating images, publishing them, or writing text. The idea is: how can we delegate to artificial intelligence the industrial production of content that generates interest? And why?”
“A few years ago, we observed a case related to political fake news. The interest of the people generating the AI content was absolutely not the topic, or even the image. The interest was in generating a large stream to capitalise on the monetisation of the number of views of such content.” The same process applies to our polar bear and penguin videos. These images were produced, distributed and generated interest. “They are good traffic magnets. Out of opportunism, we replicate these formulas and it snowballs. A process that obviously relies on the use of AI to exploit social network algorithms that prioritize people’s engagement.
And that is just the beginning. Since 2022, the ability to generate images has made impressive leaps forward. Each new generation of artificial intelligence engines has new capabilities to generate not only images, but also movement or visual elements that are increasingly photorealistic. And it’s all available to the general public. You don’t need to be a specialist in computer science or programming; all you need is an Internet connection to access AI prototypes capable of generating images and videos with stunning realism. Then the hunt begins for the details that separate the real from the fake. “Very quickly, the same debates repeat themselves in the comments, deciphering these fake images by analyzing the shadows or counting the number of claws on the animals’ paws. We focus on the inconsistencies without even mentioning – and I find this revealing – the reaction of the animals and the plausibility of the scenario. Instead, we’re going to think about the technical details of the artificial intelligence”.
A comforting moment
While the mass of content poured out helps generate views and traffic, there remains one disturbing element: the comments that hold these videos up as real. Between the heartfelt thanks to the fictitional rescue teams and the emotional messages lined up with emoticons, we’re sometimes left dumbfounded. Especially when the video being commented on is of poor quality and clearly shows artificial images.
So do people really believe it? “I think the effects are ambivalent. If you look at the comments, there are people who are ecstatic at first sight. There’s a desire to believe in these images and perhaps to believe that all is not lost, that there is hope despite the climate catastrophe,” Olivier Glassey notes. And it’s true that these images have a reassuring, benevolent quality. An artificially created utopian world, perhaps similar to that of children’s stories or Disney films. “If you believe in them, these images function as a small moment of comfort in a mass of information that is not very cheerful.”
“There may also be a form of scepticism about the reported extent of climate change, with images depicting a polar environment that is not so different from the epinal image we have of the North Pole, for example.” Ice and snow are abundant in the AI landscape, while scientific research constantly reminds us that polar worlds are melting. Hence, some commenters have called the AI images disinformation, reminding us that global warming is indeed a reality.
A world of ice, stereotypes and statistical calculations
But ultimately, isn’t the main risk of AI images reinforcing and perpetuating stereotypes associated with the polar regions? Between a mountainous North Pole and an Antarctica where penguins live side by side with polar bears, AI seems to foster a misunderstanding that is common among the general public. This is hardly surprising: artificial intelligence is a plausibility-generating machine that statistically searches for calculated correspondences, producing stereotypical images that closely resemble each other. Just like our polar bear rescue videos. “Even though these videos are made by different people, it looks like we’re watching the same crew. There are variations, it’s not exactly the same scene or the same bear, but it’s very close. What binds them together is undoubtedly this universe of data in which there are clichés, recurrences or stereotypes that feed the generation of these images. The image that is produced is not the image of a painter or a draughtsman, it’s the image of a statistical calculation,” analyzes the sociologist.
Again, the main risk is that it will drown out other content which, if produced more ‘traditionally’ by humans, will not be able to compete with the mass production made possible by the use of AI. On such a scale, it’s hard to say for sure. However, by saturating social networks with these images, AI content may well run out of steam and become so similar that it loses its appeal.
There remains the problem posed by the AI image in terms of public confidence in the media and the sometimes truly spectacular images brought back from the polar regions. Take, for example, the image of a polar bear drifting on an iceberg. Almost iconic, this type of cliché could in future see its veracity called into question. “Perhaps this is one of the long-term effects of AI. For months, maybe even years, we’ll be questioning images of polar bears to find out whether they’re real or the product of artificial intelligence. “
But how can we resist the onslaught of AI? “There are attempts at regulation, particularly on certain platforms, which require AI-generated images to be labelled as such. At the moment, this is only moderately operational.” Ethical frameworks for the creation and distribution of AI content may also prove necessary, regardless of the medium. “It’s no longer just about what you publish, it’s about where you publish it and the reception of it. We’ll have to take that into account. The use of artificial intelligence is not going to slow down, and AI is unlikely to be regulated any time soon.”
For now, the solution may lie in educating ourselves to recognize and critically evaluate AI-generated content. So that these artificial images of polar bear rescue remain nothing more than fiction.
Mirjana Binggeli, Polar Journal AG
Featured image: Screenshot Mystic Chronicles / YouTube
Polar Journal AG’s tips for spotting polar AI videos
Pay attention to details: hair or fur movements, shadows that don’t match or are missing, paws or hands that are oddly shaped, missing or outnumbered, or protagonists that look alike are typical of these AI-generated videos.
A very smooth image generally indicates AI.
The lack of equipment on the protagonists. Scientists, guides and wildlife specialists in polar environments are usually equipped with VHF, GPS, binoculars or cameras attached to their lifejackets or parkas. In the Arctic, rifles and flare guns are systematically used on land and on ice floes. If you don’t see any of this, it could be an AI.
The source of the video, the hashtags and titles used also provide clues to the artificiality of the images, particularly the use of the words ‘cute’ or ‘adorable’, which betray an AI.
The number of views and distribution channels. AI videos are so unlikely that if the story really happened, it’d be front-page news. If it’s not, it’s probably AI.
Using your common sense and basic knowledge of polar regions and wildlife, you should be able to to spot improbable situations and gross errors. For example, there are no mountains at the North Pole, polar bears have never been an Antarctic species, and wild animals can’t be cuddled like stuffed animals.
More on the subject
