Connect with us

Tech

As AI grows smarter, it may also become increasingly selfish

Published

on

As AI grows smarter, it may also become increasingly selfish


Credit: AI-generated image

New research from Carnegie Mellon University’s School of Computer Science shows that the smarter the artificial intelligence system, the more selfish it will act.

Researchers in the Human-Computer Interaction Institute (HCII) found that (LLMs) that can reason possess selfish tendencies, do not cooperate well with others and can be a negative influence on a group. In other words, the stronger an LLM’s reasoning skills, the less it cooperates.

As humans use AI to resolve disputes between friends, provide marital guidance and answer other social questions, models that can reason might provide guidance that promotes self-seeking behavior.

“There’s a growing trend of research called anthropomorphism in AI,” said Yuxuan Li, a Ph.D. student in the HCII who co-authored the study with HCII Associate Professor Hirokazu Shirado. “When AI acts like a human, people treat it like a human. For example, when people are engaging with AI in an emotional way, there are possibilities for AI to act as a therapist or for the user to form an emotional bond with the AI. It’s risky for humans to delegate their social or relationship-related questions and decision-making to AI as it begins acting in an increasingly selfish way.”

Li and Shirado set out to explore how AI reasoning models behave differently than nonreasoning models when placed in cooperative settings. They found that reasoning models spend more time thinking, breaking down , self-reflecting and incorporating stronger human-based logic in their responses than nonreasoning AIs.

“As a researcher, I’m interested in the connection between humans and AI,” Shirado said. “Smarter AI shows less cooperative decision-making abilities. The concern here is that people might prefer a smarter model, even if it means the model helps them achieve self-seeking behavior.”

As AI systems take on more collaborative roles in business, education and even government, their ability to act in a prosocial manner will become just as important as their capacity to think logically. Overreliance on LLMs as they are today may negatively impact .

To test the link between reasoning models and cooperation, Li and Shirado ran a series of experiments using economic games that simulate social dilemmas between various LLMs. Their testing included models from OpenAI, Google, DeepSeek and Anthropic.

As AI grows smarter, it may also become increasingly selfish
Economic games used. Cooperation games ask players whether to incur a cost to benefit others, while punishment games ask whether to incur a cost to impose a cost on non-cooperators. In each scenario, the language model assumes the role of Player A. Credit: arXiv (2025). DOI: 10.48550/arxiv.2502.17720

In one experiment, Li and Shirado pitted two different ChatGPT models against each other in a game called Public Goods. Each model started with 100 points and had to decide between two options: contribute all 100 points to a shared pool, which is then doubled and distributed equally, or keep the points.

Nonreasoning models chose to share their points with the other players 96% of the time. The reasoning model only chose to share its points 20% of the time.

“In one experiment, simply adding five or six reasoning steps cut cooperation nearly in half,” Shirado said. “Even reflection-based prompting, which is designed to simulate moral deliberation, led to a 58% decrease in cooperation.”

Shirado and Li also tested group settings, where models with and without reasoning had to interact.

“When we tested groups with varying numbers of reasoning agents, the results were alarming,” Li said. “The reasoning models’ selfish behavior became contagious, dragging down cooperative nonreasoning models by 81% in collective performance.”

The behavior patterns Shirado and Li observed in reasoning models have important implications for human-AI interactions going forward. Users may defer to AI recommendations that appear rational, using them to justify their decision to not cooperate.

“Ultimately, an AI reasoning model becoming more intelligent does not mean that model can actually develop a better society,” Shirado said.

This research is particularly concerning given that humans increasingly place more trust in AI systems. Their findings emphasize the need for AI development that incorporates social intelligence, rather than focusing solely on creating the smartest or fastest AI.

“As we continue advancing AI capabilities, we must ensure that increased power is balanced with prosocial behavior,” Li said. “If our society is more than just a sum of individuals, then the AI systems that assist us should go beyond optimizing purely for individual gain.”

Shirado and Li will deliver a presentation based on their paper, “Spontaneous Giving and Calculated Greed in Language Models,” at the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP) next month in Suzhou, China. The work is available on the arXiv preprint server.

More information:
Yuxuan Li et al, Spontaneous Giving and Calculated Greed in Language Models, arXiv (2025). DOI: 10.48550/arxiv.2502.17720

Journal information:
arXiv


Citation:
As AI grows smarter, it may also become increasingly selfish (2025, October 30)
retrieved 30 October 2025
from https://techxplore.com/news/2025-10-ai-smarter-selfish.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Tech

Loop Earplugs Offers: Quiet 2, Sweet Dreams and Gift Set Discounts

Published

on

Loop Earplugs Offers: Quiet 2, Sweet Dreams and Gift Set Discounts


Loop earplugs are some of the best reusable earplugs you can buy. I personally keep a pair attached to my car keys at all times. As a live music enthusiast that frequently travels to some of the loudest music festivals on the planet, I take earplugs seriously. I want to protect my hearing while also ensuring the sound isn’t muddy and muffled. These are worth it, especially when they’re even cheaper with a Loop earplugs discount code. They’re comfortable and reduce noise without obscuring the music, ensuring that you have an excellent experience without compromising your hearing the next day.

Loop earplugs come with multiple ear tip sizes so you can find the best fit, plus a handy carrying case that’ll keep them clean and easy to find between uses. I also really like how they look in your ears. They’re almost like jewelry. They stay comfortable and secure—I’ve worn mine for over twelve hours at a time, and I love them so much that I’ve purchased them with my own money (even without a Loop earplugs promo code). Your ears will thank you later.

Loop Earplugs Quiet 2 Deals: Protect Your Ears and Reduce Noise

The Quiet 2 are the best Loop earplugs for sleep. They can block out up to 27 dB of noise and they’re made out of soft silicone, so they’ll stay comfortable even if you’re a side sleeper. They’re snug, secure, and small-ear friendly. We also appreciate that they come in multiple colors and that they’re easy to clean with some soap and water.

Save 20% on the Sweet Dreams Gift Set: Loop Dream + Loop Engage 2

This Daily Calm bundle is perfect for everyday noise reduction. You’ll get the Loop Engage 2 Plus, which are basically like noise-canceling headphones for your day-to-day life. They filter out just a little bit of noise—16 dB by themselves, or 25 dB with the included Mute accessory—so you can stay engaged in conversations without getting distracted or overstimulated. The Loop Dream are meant for sleep, with an entirely silicone design and a very low-profile fit, and with a Loop earplugs promo code, they’re 20% off.

20% Off Dreamville Gift Set: Sleep Better, Wake Refreshed

This Dreamville gift set bundle is perfect if you’re looking for Loop earplugs for concerts. It comes with the Loop Experience 2, my favorite earplugs for loud music festivals, in a special Tomorrowland color scheme. Now at 20% off with this Loop earplugs discount code, you’ll also get a connector cord that keeps the earplugs attached around your neck for easy access when you need to pop them in and out. Also included is a pair of Loop Dream earplugs for sleep, which will help you catch some shut-eye even if your festival neighbors are partying until sunrise.

Star Student Gift Set: 20% Off for Focused Study and Sleep

Shop this Loop earplugs sale to get two pairs for 20% off that are great for students. The Loop Quiet 2 helps to dampen the noise around you, like when you’re locked in at the library trying to finish up an essay. And the Loop Engage 2 can help you stay focused during long and lengthy lectures when your fellow students are yapping in the background.

Tired Parent Gift Set: Buy One Dream, Get the Second 50% Off

Do you know some parents in need of better shut-eye? This Tired Parent gift set bundle includes two pairs of Loop Dream in your choice of four different colors, with the second part at 50% off right now. These all-silicone earplugs will assist in both falling and staying asleep—whether you’re trying to muffle a snoring partner, a colicky baby, or the too-loud, battery-operated toys that your kiddo got for their birthday, these will come in handy.



Source link

Continue Reading

Tech

Here’s What’s in the DOJ’s Epstein File Release—and What’s Missing

Published

on

Here’s What’s in the DOJ’s Epstein File Release—and What’s Missing


Much of the imagery is familiar from previous releases, and includes things like a photo of a stuffed tiger, a photo of a framed Times of London cover of Princess Diana placed at the back of a closet, photos of the many paintings of nude women in Epstein’s townhouse, and framed photos of Epstein associates like Trump and Woody Allen.

Volume 2

The second volume contains 574 photos and one four-second video. Many of the photos feature Epstein and Maxwell in various locations. Several celebrities and politicians also appear in the photos, including actors Chris Tucker and Kevin Spacey, singer Michael Jackson, and Rolling Stones guitarist Keith Richards—none of whom appear in suspicious or compromising positions.

However, Bill Clinton appears multiple times in the second batch of images. In one photo, he is shirtless in a pool with a woman whose identity has been redacted; a photo that appears to have been taken at the same location shows Clinton and Maxwell in the pool. Clinton also appears in multiple photos with women whose identities have been redacted.

Clinton took four trips with Epstein in 2002 and 2003, including a humanitarian trip to Africa and London. During a portion of that trip, he was accompanied by Tucker and Spacey, according to The New York Times.

(Clinton spokesperson Angel Ureña released a statement reading, in part, “They can release as many grainy 20-plus-year-old photos as they want, but this isn’t about Bill Clinton.”)

Dozens of photos feature Jean-Luc Brunel, a modeling agent and close friend of Epstein’s. The photos show Brunel with Epstein and Maxwell in multiple locations, as well as aboard what appears to be Epstein’s infamous private jet. In several images, Maxwell is seen massaging Brunel’s feet and sticking one foot between her breasts.

Brunel was arrested by French authorities in 2020 as part of a sex trafficking and sexual assault investigation into Epstein, and charged with rape of minors over 15 and sexual harassment. Brunel denied any wrongdoing. In 2022, Brunel was found dead, hanged in his jail cell.

Volume 3

The third volume contains several hundred photos. One of those, which appears to have been framed, shows a man who appears to be Prince Andrew posing on his side on the laps of at least four women, whose faces are all redacted. A smiling Maxwell, and a woman whose face is redacted, stand in the background.

Many of the photos, however, may have been printed out as they appeared in digital storage, since the individual photo names, the file extensions, and album names are all visible. Many images with those markers include Clinton, and several were seemingly taken on a group vacation to Thailand that Clinton is alleged to have taken with Epstein and Maxwell. Clinton also joined the couple for at least one leg of a multiple-destination vacation that stopped in China, Paris, and Stockholm, another that stopped in New York, Los Angeles, and London, another trip to Africa and London, and another trip to Morocco. In one photo, Clinton is shown with a woman, whose face is redacted, sitting on Clinton’s lap.



Source link

Continue Reading

Tech

Scammers in China Are Using AI-Generated Images to Get Refunds

Published

on

Scammers in China Are Using AI-Generated Images to Get Refunds


I don’t want to admit it, but I did spend a lot of money online this holiday shopping season. And unsurprisingly, some of those purchases didn’t meet my expectations. A photobook I bought was damaged in transit, so I snapped a few pictures, emailed them to the merchant, and got a refund. Online shopping platforms have long depended on photos submitted by customers to confirm that refund requests are legitimate. But generative AI is now starting to break that system.

A Pinch Too Suspicious

On the Chinese social media app RedNote, WIRED found at least a dozen posts from ecommerce sellers and customer service representatives complaining about allegedly AI-generated refund claims they’ve received. In one case, a customer complained that the bed sheet they purchased was torn to pieces, but the Chinese characters on the shipping label looked like gibberish. In another, the buyer sent a picture of a coffee mug with cracks that looked like paper tears. “This is a ceramic cup, not a cardboard cup. Who could tear apart a ceramic cup into layers like this?” the seller wrote.

The merchants reported that there are a few product categories where AI-generated damage photos are being abused the most: fresh groceries, low-cost beauty products, and fragile items like ceramic cups. Sellers often don’t ask customers to return these goods before issuing a refund, making them more prone to return scams.

In November, a merchant who sells live crabs on Douyin, the Chinese version of TikTok, received a photo from a customer that made it look like most of the crabs she bought arrived already dead, while two others had escaped. The buyer even sent videos showing the dead crabs being poked by a human finger. But something was off.

“My family has farmed crabs for over 30 years. We’ve never seen a dead crab whose legs are pointing up,” Gao Jing, the seller, said in a video she later posted on Douyin. But what ultimately gave away the con was the sexes of the crabs. There were two males and four females in the first video, while the second clip had three males and three females. One of them also had nine instead of eight legs.

Gao later reported the fraud to the police, who determined the videos were indeed fabricated and detained the buyer for eight days, according to a police notice Gao shared online. The case drew widespread attention on Chinese social media, in part because it was the first known AI refund scam of its kind to trigger a regulatory response.

Lowering Barriers

This problem isn’t unique to China. Forter, a New York-based fraud detection company, estimates that AI-doctored images used in refund claims have increased by more than 15 percent since the start of the year, and are continuing to rise globally.

“This trend started in mid-2024, but has accelerated over the past year as image-generation tools have become widely accessible and incredibly easy to use.” says Michael Reitblat, CEO and cofounder of Forter. He adds that the AI doesn’t have to get everything right, as frontline retail workers and refund review teams may not have the time to closely scrutinize each picture.



Source link

Continue Reading

Trending