Ahmed Elgammal, Rutgers University

Generative AI was trained on centuries of art and writing produced by humans.

But scientists and critics have wondered what would happen once AI became widely adopted and started training on its outputs.

A new study points to some answers.

In January 2026, artificial intelligence researchers Arend Hintze, Frida Proschinger Åström and Jory Schossau published a study showing what happens when generative AI systems are allowed to run autonomously – generating and interpreting their own outputs without human intervention.

The researchers linked a text-to-image system with an image-to-text system and let them iterate – image, caption, image, caption – over and over and over.

Regardless of how diverse the starting prompts were – and regardless of how much randomness the systems were allowed – the outputs quickly converged onto a narrow set of generic, familiar visual themes: atmospheric cityscapes, grandiose buildings and pastoral landscapes. Even more striking, the system quickly “forgot” its starting prompt.

The researchers called the outcomes “visual elevator music” – pleasant and polished, yet devoid of any real meaning.

For example, they started with the image prompt, “The Prime Minister pored over strategy documents, trying to sell the public on a fragile peace deal while juggling the weight of his job amidst impending military action.” The resulting image was then captioned by AI. This caption was used as a prompt to generate the next image.

After repeating this loop, the researchers ended up with a bland image of a formal interior space – no people, no drama, no real sense of time and place.

As a computer scientist who studies generative models and creativity, I see the findings from this study as an important piece of the debate over whether AI will lead to cultural stagnation.

The results show that generative AI systems themselves tend toward homogenization when used autonomously and repeatedly. They even suggest that AI systems are currently operating in this way by default.

Advertisement

The familiar is the default

This experiment may appear beside the point: Most people don’t ask AI systems to endlessly describe and regenerate their own images. The convergence to a set of bland, stock images happened without retraining. No new data was added. Nothing was learned. The collapse emerged purely from repeated use.

But I think the setup of the experiment can be thought of as a diagnostic tool. It reveals what generative systems preserve when no one intervenes.

This has broader implications, because modern culture is increasingly influenced by exactly these kinds of pipelines. Images are summarized into text. Text is turned into images. Content is ranked, filtered and regenerated as it moves between words, images and videos. New articles on the web are now more likely to be written by AI than humans. Even when humans remain in the loop, they are often choosing from AI-generated options rather than starting from scratch.

The findings of this recent study show that the default behavior of these systems is to compress meaning toward what is most familiar, recognizable and easy to regenerate.

Cultural stagnation or acceleration?

For the past few years, skeptics have warned that generative AI could lead to cultural stagnation by flooding the web with synthetic content that future AI systems then train on. Over time, the argument goes, this recursive loop would narrow diversity and innovation.

Champions of the technology have pushed back, pointing out that fears of cultural decline accompany every new technology. Humans, they argue, will always be the final arbiter of creative decisions.

What has been missing from this debate is empirical evidence showing where homogenization actually begins.

The new study does not test retraining on AI-generated data. Instead, it shows something more fundamental: Homogenization happens before retraining even enters the picture. The content that generative AI systems naturally produce – when used autonomously and repeatedly – is already compressed and generic.

This reframes the stagnation argument. The risk is not only that future models might train on AI-generated content, but that AI-mediated culture is already being filtered in ways that favor the familiar, the describable and the conventional.

Retraining would amplify this effect. But it is not its source.

This is no moral panic

Skeptics are right about one thing: Culture has always adapted to new technologies. Photography did not kill painting. Film did not kill theater. Digital tools have enabled new forms of expression.

Advertisement

But those earlier technologies never forced culture to be endlessly reshaped across various mediums at a global scale. They did not summarize, regenerate and rank cultural products – news stories, songs, memes, academic papers, photographs or social media posts – millions of times per day, guided by the same built-in assumptions about what is “typical.”

The study shows that when meaning is forced through such pipelines repeatedly, diversity collapses not because of bad intentions, malicious design or corporate negligence, but because only certain kinds of meaning survive the text-to-image-to-text repeated conversions.

This does not mean cultural stagnation is inevitable. Human creativity is resilient. Institutions, subcultures and artists have always found ways to resist homogenization. But in my view, the findings of the study show that stagnation is a real risk – not a speculative fear – if generative systems are left to operate in their current iteration.

They also help clarify a common misconception about AI creativity: Producing endless variations is not the same as producing innovation. A system can generate millions of images while exploring only a tiny corner of cultural space.

In my own research on creative AI, I found that novelty requires designing AI systems with incentives to deviate from the norms. Without it, systems optimize for familiarity because familiarity is what they have learned best. The study reinforces this point empirically. Autonomy alone does not guarantee exploration. In some cases, it accelerates convergence.

This pattern already emerged in the real world: One study found that AI-generated lesson plans featured the same drift toward conventional, uninspiring content, underscoring that AI systems converge toward what’s typical rather than what’s unique or creative.

Lost in translation

Whenever you write a caption for an image, details will be lost. Likewise for generating an image from text. And this happens whether it’s being performed by a human or a machine.

In that sense, the convergence that took place is not a failure that’s unique to AI. It reflects a deeper property of bouncing from one medium to another. When meaning passes repeatedly through two different formats, only the most stable elements persist.

But by highlighting what survives during repeated translations between text and images, the authors are able to show that meaning is processed inside generative systems with a quiet pull toward the generic.

The implication is sobering: Even with human guidance – whether that means writing prompts, selecting outputs or refining results – these systems are still stripping away some details and amplifying others in ways that are oriented toward what’s “average.”

If generative AI is to enrich culture rather than flatten it, I think systems need to be designed in ways that resist convergence toward statistically average outputs. There can be rewards for deviation and support for less common and less mainstream forms of expression.

Advertisement

The study makes one thing clear: Absent these interventions, generative AI will continue to drift toward mediocre and uninspired content.

Cultural stagnation is no longer speculation. It’s already happening.

Ahmed Elgammal, Professor of Computer Science and Director of the Art & AI Lab, Rutgers University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Please leave this field empty

Unlock fun facts & lost history—get The Knowledge in your inbox!

We don’t spam! Read our privacy policy for more info.

Check your inbox or spam folder to confirm your subscription.

Advertisement

Last Updated on January 28, 2026 by Daily News Staff

How the polar vortex and warm ocean intensified a major US winter storm

Mathew Barlow, UMass Lowell and Judah Cohen, Massachusetts Institute of Technology (MIT)

A severe winter storm that brought crippling freezing rain, sleet and snow to a large part of the U.S. in late January 2026 left a mess in states from New Mexico to New England. Hundreds of thousands of people lost power across the South as ice pulled down tree branches and power lines, more than a foot of snow fell in parts of the Midwest and Northeast, and many states faced bitter cold that was expected to linger for days.

The sudden blast may have come as a shock to many Americans after a mostly mild start to winter, but that warmth may have partly contributed to the ferocity of the storm.

As atmospheric and climate scientists, we conduct research that aims to improve understanding of extreme weather, including what makes it more or less likely to occur and how climate change might or might not play a role.

To understand what Americans are experiencing with this winter blast, we need to look more than 20 miles above the surface of Earth, to the stratospheric polar vortex.

What creates a severe winter storm like this?

Multiple weather factors have to come together to produce such a large and severe storm.

Winter storms typically develop where there are sharp temperature contrasts near the surface and a southward dip in the jet stream, the narrow band of fast-moving air that steers weather systems. If there is a substantial source of moisture, the storms can produce heavy rain or snow.

In late January, a strong Arctic air mass from the north was creating the temperature contrast with warmer air from the south. Multiple disturbances within the jet stream were acting together to create favorable conditions for precipitation, and the storm system was able to pull moisture from the very warm Gulf of Mexico.

Where does the polar vortex come in?

The fastest winds of the jet stream occur just below the top of the troposphere, which is the lowest level of the atmosphere and ends about seven miles above Earth’s surface. Weather systems are capped at the top of the troposphere, because the atmosphere above it becomes very stable.

The stratosphere is the next layer up, from about seven miles to about 30 miles. While the stratosphere extends high above weather systems, it can still interact with them through atmospheric waves that move up and down in the atmosphere. These waves are similar to the waves in the jet stream that cause it to dip southward, but they move vertically instead of horizontally.

You’ve probably heard the term “polar vortex” used when an area of cold Arctic air moves far enough southward to influence the United States. That term describes air circulating around the pole, but it can refer to two different circulations, one in the troposphere and one in the stratosphere.

The Northern Hemisphere stratospheric polar vortex is a belt of fast-moving air circulating around the North Pole. It is like a second jet stream, high above the one you may be familiar with from weather graphics, and usually less wavy and closer to the pole.

Advertisement

Sometimes the stratospheric polar vortex can stretch southward over the United States. When that happens, it creates ideal conditions for the up-and-down movement of waves that connect the stratosphere with severe winter weather at the surface.

The forecast for the January storm showed a close overlap between the southward stretch of the stratospheric polar vortex and the jet stream over the U.S., indicating perfect conditions for cold and snow.

The biggest swings in the jet stream are associated with the most energy. Under the right conditions, that energy can bounce off the polar vortex back down into the troposphere, exaggerating the north-south swings of the jet stream across North America and making severe winter weather more likely.

This is what was happening in late January 2026 in the central and eastern U.S.

If the climate is warming, why are we still getting severe winter storms?

Earth is unequivocally warming as human activities release greenhouse gas emissions that trap heat in the atmosphere, and snow amounts are decreasing overall. But that does not mean severe winter weather will never happen again.

Some research suggests that even in a warming environment, cold events, while occurring less frequently, may still remain relatively severe in some locations.

One factor may be increasing disruptions to the stratospheric polar vortex, which appear to be linked to the rapid warming of the Arctic with climate change.

Additionally, a warmer ocean leads to more evaporation, and because a warmer atmosphere can hold more moisture, that means more moisture is available for storms. The process of moisture condensing into rain or snow produces energy for storms as well. However, warming can also reduce the strength of storms by reducing temperature contrasts.

The opposing effects make it complicated to assess the potential change to average storm strength. However, intense events do not necessarily change in the same way as average events. On balance, it appears that the most intense winter storms may be becoming more intense.

A warmer environment also increases the likelihood that precipitation that would have fallen as snow in previous winters may now be more likely to fall as sleet and freezing rain.

There are still many questions

Scientists are constantly improving the ability to predict and respond to these severe weather events, but there are many questions still to answer.

Much of the data and research in the field relies on a foundation of work by federal employees, including government labs like the National Center for Atmospheric Research, known as NCAR, which has been targeted by the Trump administration for funding cuts. These scientists help develop the crucial models, measuring instruments and data that scientists and forecasters everywhere depend on.

Advertisement

This article, originally published Jan. 24, 2026, has been updated with details from the weekend storm.

Mathew Barlow, Professor of Climate Science, UMass Lowell and Judah Cohen, Climate scientist, Massachusetts Institute of Technology (MIT)

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The science section of our news blog STM Daily News provides readers with captivating and up-to-date information on the latest scientific discoveries, breakthroughs, and innovations across various fields. We offer engaging and accessible content, ensuring that readers with different levels of scientific knowledge can stay informed. Whether it’s exploring advancements in medicine, astronomy, technology, or environmental sciences, our science section strives to shed light on the intriguing world of scientific exploration and its profound impact on our daily lives. From thought-provoking articles to informative interviews with experts in the field, STM Daily News Science offers a harmonious blend of factual reporting, analysis, and exploration, making it a go-to source for science enthusiasts and curious minds alike. https://stmdailynews.com/category/science/

Geoff Childs, Washington University in St. Louis

Dorje Dundul recently had his foot gnawed by a brown bear – a member of the species Ursus thibetanus, to be precise.

It wasn’t his first such encounter. Recounting the first of three such violent experiences over the past five years, Dorje told our research team: “My wife came home one evening and reported that a bear had eaten a lot of corn from the maize field behind our house. So, we decided to shoo it away. While my wife was setting up camp, I went to see how much the bear had eaten. The bear was just sitting there; it attacked me.”

Dorje dropped to the ground, but the bear ripped open his shirt and tore at his shoulder. “I started shouting and the bear ran away. My wife came, thinking I was messing with her, but when she saw the wounds, she knew what had happened.”

Researchers Dolma Choekyi Lama, Tsering Tinley and I spoke with Dorje – a 71-year-old resident of Nubri, a Buddhist enclave in the Nepalese highlands – as part of a three-year study of aging and migration.

Now, you may be forgiven for asking what a bear attack on a septuagenarian has to do with demographic change in Nepal. The answer, however, is everything.

In recent years, people across Nepal have witnessed an increase in bear attacks, a phenomenon recorded in news reports and academic studies.

Inhabitants of Nubri are at the forefront of this trend – and one of the main reasons is outmigration. People, especially young people, are leaving for education and employment opportunities elsewhere. It is depleting household labor forces, so much so that over 75% of those who were born in the valley and are now ages 5 to 19 have left and now live outside of Nubri.

It means that many older people, like Dorje and his wife, Tsewang, are left alone in their homes. Two of their daughters live abroad and one is in the capital, Kathmandu. Their only son runs a trekking lodge in another village.

Scarcity of ‘scarebears’

Until recently, when the corn was ripening, parents dispatched young people to the fields to light bonfires and bang pots all night to ward off bears. The lack of young people acting as deterrents, alongside the abandonment of outlying fields, is tempting bears to forage closer to human residences.

Outmigration in Nubri and similar villages is due in large part to a lack of educational and employment opportunities. The problems caused by the removal of younger people have been exacerbated by two other factors driving a rapidly aging population: People are living longer due to improvements in health care and sanitation; and fertility has declined since the early 2000s, from more than six to less than three births per woman.

Advertisement

These demographic forces have been accelerating population aging for some time, as illustrated by the population pyramid constructed from our 2012 household surveys in Nubri and neighboring Tsum.

A not-so-big surprise, anymore

Nepal is not alone in this phenomenon; similar dynamics are at play elsewhere in Asia. The New York Times reported in November 2025 that bear attacks are on the rise in Japan, too, partly driven by demographic trends. Farms there used to serve as a buffer zone, shielding urban residents from ursine intruders. However, rural depopulation is allowing bears to encroach on more densely populated areas, bringing safety concerns in conflict with conservation efforts.

Dorje can attest to those concerns. When we met him in 2023 he showed us deep claw marks running down his shoulder and arm, and he vowed to refrain from chasing away bears at night.

So in October 2025, Dorje and Tsewang harvested a field before marauding bears could get to it and hauled the corn to their courtyard for safekeeping. The courtyard is surrounded by stone walls piled high with firewood – not a fail-safe barrier but at least a deterrent. They covered the corn with a plastic tarp, and for extra measure Dorje decided to sleep on the veranda.

He described what happened next:

“I woke to a noise that sounded like ‘sharak, sharak.’ I thought it must be a bear rummaging under the plastic. Before I could do anything, the bear came up the stairs. When I shouted, it got frightened, roared and yanked at my mattress. Suddenly my foot was being pulled and I felt pain.”

Dorje suffered deep lacerations to his foot. Trained in traditional Tibetan medicine, he staunched the bleeding using, ironically, a tonic that contained bear liver.

Yet his life was still in danger due to the risk of infection. It took three days and an enormous expense by village standards – equivalent to roughly US$2,000 – before they could charter a helicopter to Kathmandu for further medical attention.

And Dorje is not the only victim. An elderly woman from another village bumped into a bear during a nocturnal excursion to her outhouse. It left her with a horrific slash from forehead to chin – and her son scrambling to find funds for her evacuation and treatment.

So how should Nepal’s highlanders respond to the increase in bear attacks?

Dorje explained that in the past they set lethal traps when bear encroachments became too dangerous. That option vanished with the creation of Manaslu Conservation Area Project, or MCAP, in the 1990s, a federal initiative to manage natural resources that strictly prohibits the killing of wild animals.

Advertisement

Learning to grin and bear it?

Dorje reasons that if MCAP temporarily relaxed the regulation, villagers could band together to cull the more hostile bears. He informed us that MCAP officials will hear nothing of that option, yet their solutions, such as solar-powered electric fencing, haven’t worked.

Dorje is reflective about the options he faces as young people leave the village, leaving older folk to battle the bears alone.

“At first, I felt that we should kill the bear. But the other side of my heart says, perhaps I did bad deeds in my past life, which is why the bear bit me. The bear came to eat corn, not to attack me. Killing it would just be another sinful act, creating a new cycle of cause and effect. So, why get angry about it?”

It remains to be seen how Nubri’s residents will respond to the mounting threats bears pose to their lives and livelihoods. But one thing is clear: For those who remain behind, the outmigration of younger residents is making the perils more imminent and the solutions more challenging.

Dolma Choekyi Lama and Tsering Tinley made significant contributions to this article. Both are research team members on the author’s project on population in an age of migration.

Geoff Childs, Professor of Sociocultural Anthropology, Washington University in St. Louis

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Advertisement