Artificial intelligence is becoming increasingly versatile—from route planning to text translation, it has long become a standard tool. But it is not enough for AI to simply deliver useful results: it is becoming ever more important that it also complies with legal, ethical, and social norms. But how can such norms be taught to a machine?

At TU Wien, a new approach has now been developed. By combining machine learning and logic, autonomous agents can be trained to follow predefined norms. It is even possible to establish a hierarchy of these norms—declaring some to be more important than others. At IJCAI 2025, an AI conference held this year in Montreal, Canada, this work was recognized with the Distinguished Paper Award.

Trial and error

Teaching AI new abilities sometimes works a bit like teaching tricks to a pet: reward if the task is performed correctly, punishment if the response is wrong. The AI tries out different behaviors and, through trial and error, learns how to maximize its reward. This method is called reinforcement learning and plays a key role in AI research.

“One could try to teach AI certain rules by rewarding the agent for following norms. This technique works well in the case of safety constraints,” says Prof. Agata Ciabattoni from the Institute of Logic and Computation at TU Wien. “But this wouldn’t work, for instance, with conditional norms (‘do A under condition B’). If the agent finds a way to earn a reward, it might delay finishing its actual job on purpose, to have more time for scoring easy points.”

Norms as logical formulas

The TU Wien team chose a fundamentally different path, inspired by old philosophical works: norms are still represented as logical formulas, but agents get a punishment when they do not comply with them. For example, “you must not exceed the speed limit” is translated as “if you exceed the speed limit you get a punishment of X.” Most importantly, each norm is treated as an independent objective.

“The artificial agent is given a goal to pursue—for example, to find the best route to a list of destinations. At the same time, we also define additional rules and norms that it must observe along the way,” explains Emery Neufeld, the first author of the paper. “The fact that each norm is treated as a different objective allows us to algorithmically compute the relative weight that we have to assign to these objectives in order to get a good overall result.”

With this technique, it becomes possible to encode even complicated sets of rules—for instance, norms that apply only under certain conditions, or norms that depend on the violation of other norms.

Flexible norms

“The great thing is that when the norms change, the training does not have to start all over again,” says Agata Ciabattoni. “We have a system that learns to comply with norms—but we can then still adjust these norms afterwards, or change their relative importance, declaring one rule to be more important than another.”

In their paper, Ciabattoni and her team were able to show that this technique allows a wide range of norms to be imposed, while the AI continues to pursue its primary goals.

A new study in Acta Psychologica finds that South Korea’s new law requiring mobile games to disclose loot box probabilities is more effective than industry self-regulation.

Researchers Leon Y. Xiao (City University of Hong Kong & beClaws.org) and Solip Park (Aalto University) analyzed the 100 top-grossing iPhone games in South Korea.

They found that:

• 90% of games included paid loot boxes.
• 84% of games with loot boxes disclosed probabilities—compared to just 35% in the Netherlands and 64% in the UK.
• Only 41% of disclosures were easy to find, meaning many players may fail to access the information and remain confused or uninformed.

South Korean regulators have backed the law with active enforcement, identifying noncompliance and forcing companies to fix mistakes. Separately, fines were levied against major South Korean publishers like Nexon for misleading disclosures.

“South Korea is showing the world that loot box regulation can work—but only when actively enforced,” said Xiao. “Industry self-regulation has failed globally. If governments really want to protect players, we need enforceable laws with real penalties.”

The authors call on other countries to follow South Korea’s lead in actively enforcing video game regulations, but also to strengthen standards so that disclosures are clear, accessible, and independently evaluated for accuracy.

For most people, though, “if results are strong, maybe you don’t need another scan for five years,” says Wagner. “If they’re lower, lifestyle interventions can help, and you may want to recheck in a year.”

Radiation exposure is negligible, less than a chest x-ray. But the psychological impact can be more complicated. For some, the numbers motivate: “When I did a body composition test at 36, I had way more body fat than I expected,” Cheema says. “That pushed me to change my workouts and eating patterns in ways that improved my health—something BMI alone wouldn’t have prompted.”

For others, especially those with histories of disordered eating or body image issues, it can be destabilizing and overwhelming. Numbers can become another metric to obsess over rather than a tool for health. “It can be overwhelming if you don’t have a clinician to interpret the results,” Gidwani says. “That’s why I review all of my patients’ scans with them.”

Cheema agrees: “Too much detail without guidance risks overwhelming people with information that isn’t clinically actionable.”

“I don’t think DEXA gives too much information compared to, say, a whole-body MRI, which can reveal incidental findings that can cause anxiety and lead to unnecessary interventions,” says Gidwani. “Its data points are actionable: decrease body fat, reduce visceral fat, increase muscle.”

Experts emphasize that actionability is key. “The most important metrics are visceral adipose tissue and total body fat percentage, especially when tracked over time,” Cheema says. “But DEXA also breaks things down by arms, legs, trunk, etc. That can veer into aesthetics rather than health.

Should You Get One?

If you’re 65 or older, or at risk for osteoporosis, your doctor may already recommend a DEXA scan for bone health. For women in perimenopause, when bone density can drop by as much as 20 percent, an early baseline scan could flag risks years before they become urgent.

DEXA also detects sarcopenic obesity, where muscle loss occurs alongside high body fat. “Someone may look normal weight on a scale, but a DEXA can reveal poor muscle-to-fat balance,” Gidwani says.

Beyond those groups, the use case narrows. Athletes, bodybuilders, and people on GLP-1 medications may find the data genuinely useful. For generally healthy adults who exercise, eat decently, and check in with a doctor, many clinicians are indifferent.

“For a healthy individual, I wouldn’t universally recommend it,” Cheema says. “Lifestyle changes and basic care may matter more than getting a DEXA.” There are alternatives—bioimpedance scales, Bod Pods, and AI-enabled wearables—but none are as accurate as DEXA. For now, it remains the most precise, if expensive, tool available.

Final Takeaways

My DEXA results were somewhat humbling. Despite near-daily workouts and a decent diet, the scan flagged more body fat than I expected and the beginnings of osteopenia in my spine. The bright side was an “excellent” visceral fat score, something I’ll be bragging about indefinitely.

Catching early bone loss feels actionable; I can tweak my workouts to prioritize strength and mobility. But the body fat percentages have lived in my brain rent-free ever since, without offering much in return. I don’t plan to shell out a few hundred dollars for another scan anytime soon, so I may never know if my adjustments are actually working.

That’s the paradox of DEXA. For those with medical risks, it can be invaluable. For athletes chasing marginal gains, it’s another knob to turn. But for the rest of us, it’s a reminder that data is only as useful as what you’re willing or able to do with it. In the end, DEXA doesn’t promise longevity so much as it promises numbers, and numbers alone don’t add years to your life.

Meet the Experts

• Jennifer Wagner, MD, MS, chief health and performance officer, Canyon Ranch in Tucson, Arizona.
• Josh Cheema, MD, medical director of Northwestern Medicine Human Longevity Clinic in Chicago, Illinois.
• Pooja Gidwani, MD, MBA, board-certified physician in internal medicine and obesity medicine in Los Angeles, California.