Monday, August 18, 2025

Rain halts rescue operation after Pakistan floods kill hundreds


By AFP
August 18, 2025


Flash floods killed more than 200 people in worst-hit Buner district of Pakistan's Khyber Pakhtunkhwa province - Copyright AFP Abdul MAJEED
Ahmad KHAN with Lehaz ALI in Peshawar

Rain on Monday halted search and rescue operations in northern Pakistan after flash floods that have killed nearly 350 people with around 200 still missing, officials said.

Torrential rains across the country since Thursday have caused flooding, rising waters and landslides that have swept away entire villages and left many residents trapped in the rubble.

Most of the deaths — more than 320 — were reported in northwestern Khyber Pakhtunkhwa province, according to the provincial disaster agency, which warned of new flash floods “till Thursday”.

Volunteers had been assisting hundreds of rescue workers in their race against time to find possible survivors and retrieve bodies as fresh rains started lashing the province.

“This morning fresh rains forced a halt to relief operations,” said Nisar Ahmad, 31, a volunteer in worst-hit Buner district, where “12 villages have been completely wiped out and 219 bodies have been recovered”.

“Dozens of bodies are still buried under the mud and rocks, which can only be recovered with heavy machinery. However, the makeshift tracks built to access the area have once again been destroyed by the new rains,” he added.

Around 200 people are still missing in Khyber Pakhtunkhwa province, according to local authorities.



– ‘We feel scared’ –



Many people fled to seek shelter under damaged infrastructure and in the mountains in Buner, an area with difficult terrain.

“Even if it rains a little now, we feel scared because there was light rain that day. And then the unsuspecting people were swept away by the storm,” said Buner resident Ghulam Hussain, 35.

“Children and women are running and screaming up the mountains to escape,” Hazrat Ullah, 18, told AFP.

Volunteer Ahmad said there were also fears for the future due to a lack of food supplies and clean water.

“Many livestock have also perished in the cloudburst, and their decomposing bodies are spreading a foul odour in several places. Right now, our most urgent need is clean drinking water, and I appeal to the government to provide it,” he said.

The monsoon season brings about three-quarters of South Asia’s annual rainfall, which is vital for agriculture and food security but also causes widespread destruction.

According to the National Disaster Agency, the intensity of this year’s monsoon is about 50 to 60 percent higher than last year.

Landslides and flash floods are common during the monsoon season, which typically begins in June and lasts until the end of September.

The heavy rains that have battered Pakistan since the start of the summer monsoon have claimed the lives of more than 650 people, with over 920 injured.

Pakistan is among the world’s most vulnerable countries to the effects of climate change and is increasingly facing extreme weather events.

In 2022, monsoon floods submerged one-third of the country and resulted in approximately 1,700 deaths.


Hopes for survivors wane as landslides, flooding bury Pakistan villages

By AFP
August 17, 2025


People gather near a damaged vehicle and debris after flash flooding wiped out a road in monsoon-hit northern Pakistan - Copyright AFP Mehboob UL HAQ
Lehaz Ali

Thousands of Pakistani rescuers battled rain and knee-deep mud Sunday, digging homes out from under massive boulders in a desperate search for survivors after flash floods killed at least 344 people in the country’s mountainous north.

Most of the deaths were reported in Khyber Pakhtunkhwa province, where monsoon rains that are only expected to intensify in the days ahead drove flooding and landslides that collapsed houses.

In hardest-hit Bunar district, at least 208 people were killed and “10 to 12 entire villages” partially buried, a provincial rescue spokesman told AFP.

“The operation to rescue people trapped under debris is ongoing,” said Bilal Ahmed Faizi of Khyber Pakhtunkhwa’s rescue agency.

“There is still concern that dozens of people may be trapped under the rubble… the chances of those buried under the debris surviving are very slim”.

He said that around 2,000 rescue workers were engaged in recovering bodies from the debris and carrying out relief operations across nine districts, where rain was still hampering efforts.

AFP journalists in Buner saw half-buried vehicles and belongings lying strewn in the sludge, with mud covering houses and shops.

Flooded roads hampered the movement of rescue vehicles, as a few villagers worked to cut fallen trees to clear the way after the water receded.

“Our belongings are scattered, ruined and are in bad shape,” local shopkeeper Noor Muhammad told AFP as he used a shovel to remove mud.

“The shops have been destroyed along with everything else. Even the little money people had has been washed away,” he added.

The provincial government has declared the severely affected mountainous districts of Buner, Bajaur, Swat, Shangla, Mansehra and Battagram as disaster-hit areas.

“This disaster has spread everywhere and surrounded us from all sides. We were trapped in our homes and could not get out, another Buner resident, Syed Wahab Bacha, told AFP.

“Our entire poor community has been affected. The shops in the lower bazaar have been destroyed. This road was our only path, and it too has been washed away,” he added.



– Mass funerals –



On Saturday, hundreds of locals gathered for mass funerals, where bodies wrapped in blood-stained white shawls were laid out on the village ground.

Fallen trees and straw debris were scattered across nearby fields, while residents shoveled mud brought in by the floods out of their homes.

Pakistan’s meteorological department has forecast that “torrential rains” with monsoon activity were “likely to intensify” from Sunday onwards.

The department warned of more flash floods and landslides in the country’s northwest and urged people to avoid exposure to vulnerable areas.

The monsoon season brings South Asia about three-quarters of its annual rainfall, vital for agriculture and food security, but also brings destruction.

Landslides and flash floods are common during the season, which usually begins in June and eases by the end of September.

The national disaster agency’s Syed Muhammad Tayyab Shah told AFP that this year’s monsoon season began earlier than usual and was expected to end later.

The torrential rains that have pounded Pakistan since the start of the summer monsoon, described as “unusual” by authorities, have killed at least 650 people, with more than 910 injured.

Pakistan is one of the world’s most vulnerable countries to the effects of climate change and is contending with extreme weather events with increasing frequency.

Monsoon floods in 2022 submerged a third of the country and killed around 1,700 people.

Another villager in Buner told AFP on Saturday that residents had spent the night searching through the rubble of their former homes.

“The entire area is reeling from profound trauma,” said 32-year-old schoolteacher Saifullah Khan.

“I helped retrieve the bodies of the children I taught. I keep wondering what kind of trial nature has imposed on these kids,” he said.





Decision-making by people: How this improves teamwork and cross-cultural relations


By Dr. Tim Sandle
August 17, 2025
DIGITAL JOURNAL


Many exiled Tibetans accuse China's ruling Communist Party of repression and eroding their culture - Copyright AFP Santiago MAZZAROVICH

When it comes to the crunch, we are happy to ignore artificial intelligence and resort to our own reasoning, according to a new Canadian study. A study has revealed that when faced with complex decisions, people across cultures—from bustling megacities to remote Amazon communities—tend to rely on their own judgment rather than seeking advice.

Researchers from the University of Waterloo sought to learn how people from diverse cultures make decisions. This work is described as the broadest test of decision-style preferences across cultures to date.

The research, spanning over 3,500 participants in 12 countries, challenges our reliance on advances in machine learning as well as the long-held belief that self-reliance is primarily a Western trait.

Cultural differences

While cultural values influence how strongly individuals lean on their inner voice, the preference for private reflection remains a shared human tendency.

The researchers observe that by understanding that even in interdependent societies most people prefer to go with the decision made by themselves, irrespective of what others say, can help clarify cross-cultural misunderstandings and realize that we all appear to be juggling similar internal debates.

According to lead researcher Dr. Igor Grossmann, professor in the Department of Psychology at Waterloo: “Realizing that most of us instinctively ‘go it alone’ helps explain why we often ignore good counsel, be it for health tips or financial planning, despite mounting evidence: that such counsel may help us make wiser decisions…This knowledge can help us design teamwork better by working with this self-reliant tendency and letting employees reason privately before sharing advice that they might reject.”

In essence, the study upends the belief that westerners work things out themselves while the rest of the world leans on others, as does artificial intelligence.

In fact, intuition and self-reflection beat out advice from friends or crowdsourcing in all countries studied. The amount of that preference varied, depending on the level at which a culture values independence or interdependence.

Grossmann adds: “Our take-home message is that we all look inward first, yet the wisest moves may happen when solo reflections are shared with others… What culture does is controls the volume knob, dialling up that inner voice in highly independent societies and softening it somewhat in more interdependent ones.”

What does this mean for AI?

The same appears true with AI. While AI is increasingly used to assist and automate decision-making, humans still make the ultimate choices, especially in critical situations. Many experts believe that the most effective approach is a “human-in-the-loop” model, where AI provides support, and humans make the final decisions.

The research appears in the journal Proceedings of the Royal Society B Biological Science, titled “Decision-making preferences for intuition, deliberation, friends or crowds in independent and interdependent societies.”

Rooms of their own: women-only communities thrive in China

By AFP
August 17, 2025


Chen Yani (R) eats lunch with friends and guests at her women's co-living space 'Keke's Imaginative Space' in Hangzhou - Copyright AFP Jade GAO
Ludovic EHRET

Laughter erupts over a board game and coffee at a rural cottage in China’s eastern province of Zhejiang, one of a growing number of women-only co-living spaces far from social pressures and male judgment.

Women come to share mutual support and “talk freely about intimate stuff” while others seek companionship or refuge from harassment, participants told AFP after making steamed buns in a bright kitchen overlooking the mountains.

“An all-women environment makes me feel safe,” said Zhang Wenjing, 43.

“Among women, we talk more easily about certain things,” she added.

Chen Fangyan, 28, said she felt less self-conscious without men around.

“Not being forced to wear a bra is already a kind of freedom.”

Demand for single-gender spaces — including bars, gyms, hostels and co-working hubs — has grown in China, as women flex increasing economic power to secure peace of mind and physical safety.

At “Keke’s Imaginative Space”, participants pay 30 yuan ($4.17) a night, with costs going up to 80 yuan from the fourth day.

Founder Chen Yani, nicknamed “Keke”, told AFP she was motivated to open the space after bad experiences with men in the workplace.

“I encountered various degrees of harassment from men, to the point where I often found myself unable to work normally,” the 30-year-old said.

“I started thinking about what a safe and relaxed work environment would look like… a place where I wouldn’t feel apprehensive.”

– ‘Just be themselves’ –

Chen started by renovating a house in Lin’an, a suburb of Hangzhou, roughly 200 kilometres (124 miles) from Shanghai.

Believing that other women might share her desire for somewhere they could feel at ease, she organised a stay over Chinese New Year on the Instagram-like Xiaohongshu, also known as Rednote.

Twelve women showed up.

Some wanted a change of scenery for the holidays, others were keen to escape intrusive questioning or pressure from relatives, including to get married and have children.

“Within the family, women often have to take care of grandparents, children and household chores. Not to mention work responsibilities,” she said. “They need a place where they don’t have to play a role and can just be themselves.”

Women’s increasing economic independence — as well as educational opportunities — means a wider scope of options, said Yuan Xiaoqian, 29, a participant.

“They can focus more on themselves… and on new needs,” she said.

Social media is also exposing women to alternative lifestyles — particularly Rednote, which offers a growing number of options for seeking community.

In Xiuxi, a village in Zhejiang, Yang Yun opened “Her Space” in June to offer women a “spiritual haven”.

With its rustic furniture and calligraphy on the walls, the property has the feel of a boutique hotel.

The idea, she said, was to ensure women always have a place to go.

“If (a woman) loses her job, her parents, has an argument with her husband, or feels exhausted by city life, she knows she can come here and find some warmth,” said Yang.

So far, 120 women have paid the 3,980-yuan membership fee to join the quickly expanding club.

“Whether they come or not is not important. The important thing is that this place exists. It gives them mental strength,” Yang said.

– Women still lack places –

Critics claim that single-gender communities foster antagonism between men and women.

At Keke’s Imaginative Space, Chen Yani denies that anyone is harbouring antipathy towards men and insists women have a right to spaces of their own.

“Women constitute a social group with shared life trajectories and problems. It’s often easier for them to understand each other and show empathy,” she said.

While she has yet to turn a profit, Chen said that was beside the point.

“As long as there’s demand, this place will continue to exist,” she said.

Founder of the all-women cultural space “Half the Sky” in Beijing, Lilith Jiang, said these community-oriented facilities fill a void.

“Men have plenty of opportunities to socialise, while drinking or while exercising,” she said.

“Women don’t have that.”

Down the line, she said non-traditional structures could offer an alternative for single women worried about ageing alone.

“Women are constantly told: ‘If you don’t get married, what will become of you when you get older?'” said Jiang.

“But long-term, all-female shared co-living spaces where women can grow old together could be a solution.”





Cross-cultural currents: Studying Wikipedia browsing habits shows people learn


By Dr. Tim Sandle
August 17, 2025
DIGITAL JOURNAL


Wikipedia has inked a partnership with Google - Copyright Sputnik/AFP Alexei Druzhinin

Wikipedia maintains a very special place on the Internet. The site features exclusively free content and no commercial advertisements. This raises the question of how much we are in charge of where our curiosity takes us in online contexts beyond Wikipedia.

Researchers have examined how nearly half a million people around the world use Wikipedia’s knowledge networks. This reveals clear differences in browsing habits between countries offering insights into cultural differences and potential drivers of curiosity and learning.

The researchers examined the habits of 482,760 Wikipedia readers from 50 different countries, using a form of information acquisition is called the “busybody.” This is a term reserved for someone who goes from one idea or piece of information to another, and the two pieces may not relate to each other much. In simpler terms, going down digital rabbit holes.

According to lead researcher, University of Pennsylvania’s Dani Bassett: “The busybody loves any and all kinds of newness, they’re happy to jump from here to there, with seemingly no rhyme or reason, and this is contrasted by the ‘hunter,’ which is a more goal-oriented, focused person who seeks to solve a problem, find a missing factor, or fill out a model of the world.”

Through this, the researchers established stark differences in browsing habits between countries with more education and gender equality versus less equality, raising key questions about the impact of culture on curiosity and learning.

Bassett recollects:

“We observed that countries that had greater inequality, in terms of gender and access to education, had people who were browsing with more intent — seeking closely related information, whereas the people in countries that had more equality were browsing expansively, with more diversity in topics — jumping from topic to topic and collecting loosely connected information.”

Adding: While we don’t know exactly why this is, we have our hunches, and we believe these findings will prove useful in helping scientists in our field better understand the nature of curiosity.”

The researchers cite three main hypotheses driving the associations between information-seeking approaches and equality.

Hypothesis one: It is possible that countries that have more inequality also have more patriarchal structures of oppression that are constraining the knowledge production approaches to be more Hunter-like. Countries that have greater equality, in contrast, are open to a diversity of ideas, and therefore a diversity of ways that we’re engaging in the world. This is more like the busybody — the one that’s moving between ideas in a very open-minded way.

Hypothesis two: A second possibility is that browsers go to Wikipedia for different purposes in different countries, citing how someone in a country with higher equality may be going to the site for entertainment or leisure rather than for work.

Hypothesis three: The third scenario contends that people in different countries who come to Wikipedia may have different ages, genders, socioeconomic status, or educational attainment, and that those differences in who’s actually coming to Wikipedia may explain the differences in the browsing patterns.

The dancer?.

One of possible finding from the study is the emergence of a third curiosity style — the “dancer,” which had previously only been hypothesized. The dancer is someone who moves along a track of information but, unlike the busybody, they make leaps between ideas in a creative, choreographed way.

Bassett says: “It’s less about randomness and more about seeing connections where others might not.”

The research features in the journal Science Advances titled “Architectural styles of curiosity in global Wikipedia mobile app readership.”


Recalibrating the moral compass: AI now meets the conditions for having free will


By Dr. Tim Sandle
August 17, 2025
DIGITAL JOURNAL


Image: — © AFP Kazuhiro NOGI

AI is advancing at a rapid speed. This has led some researchers to worry, drawing upon speculative moral questions, about the future position of machines in relation to humanity.

According to philosopher and psychology researcher Frank Martela, generative AI is close to meeting all three of the philosophical conditions of free will — the ability to have goal-directed agency, make genuine choices and to have control over its actions.

The question of whether AI can have free will is a complex philosophical and scientific debate. Current AI systems tend to operate based on algorithms and data, raising questions about whether their actions are truly self-determined or predetermined by their programming.

However, some researchers argue that AI, especially those using probabilistic methods or exhibiting unpredictable behaviour, might possess a form of free will.

This development, Martela contends, brings us to a critical point in human history, as we give AI more power and freedom, potentially in life or death situations.

Share:

Martela’s latest study finds that generative AI meets all three of the philosophical conditions of free will — the ability to have goal-directed agency, make genuine choices and to have control over its actions.

Drawing on the concept of functional free will as explained in the theories of philosophers Daniel Dennett and Christian List, the study examined two generative AI agents powered by large language models (LLMs): the Voyager agent in Minecraft and fictional ‘Spitenik’ killer drones with the cognitive function of today’s unmanned aerial vehicles.

“Both seem to meet all three conditions of free will — for the latest generation of AI agents we need to assume they have free will if we want to understand how they work and be able to predict their behaviour,” says Martela.

He adds that these case studies are broadly applicable to currently available generative agents using LLMs.

This development brings us to a critical point in human history, as we give AI more power and freedom, potentially in life or death situations. Whether it is a self-help bot, a self-driving car or a killer drone — moral responsibility may move from the AI developer to the AI agent itself.

“We are entering new territory. The possession of free will is one of the key conditions for moral responsibility. While it is not a sufficient condition, it is one step closer to AI having moral responsibility for its actions”, Martela adds.

It follows that issues around how we ‘parent’ our AI technology have become both real and pressing.

“AI has no moral compass unless it is programmed to have one. But the more freedom you give AI, the more you need to give it a moral compass from the start. Only then will it be able to make the right choices,” Martela says.

The recent withdrawal of the latest ChatGPT update due to potentially harmful sycophantic tendencies is a red flag that deeper ethical questions must be addressed. We have moved beyond teaching the simplistic morality of a child.

Martela observes: “AI is getting closer and closer to being an adult — and it increasingly has to make decisions in the complex moral problems of the adult world. By instructing AI to behave in a certain way, developers are also passing on their own moral convictions to the AI. We need to ensure that the people developing AI have enough knowledge about moral philosophy to be able to teach them to make the right choices in difficult situations.”

The research appears in the journal AI and Ethics, titled “Artificial intelligence and free will: generative agents utilizing large language models have functional free will.”

 

Colorado State University shutters animal study after pressure from national research ethics group



Ending experiment on impact of consuming beans on gut health saves lives of more than 16,000 animals, along with hundreds of thousands of public dollars



Physicians Committee for Responsible Medicine





FORT COLLINS, Colo. — The Physicians Committee for Responsible Medicine, a nonprofit national medical ethics group, applauds Colorado State University for its decision to shutter a nutrition study for which the university had approved the killing of 17,766 animals. The study, funded by the U.S. Department of Agriculture and the university, claimed to investigate the effect of legumes on the human gut microbiome. Public records reveal the primary investigator had to date used 1,587 mice.

An initial USDA grant of $498,500 funded the experiments. A subsequent USDA Cooperative Agreement, active through Dec. 31, 2025, shared ongoing costs with the university.

Over the past two months, the Physicians Committee wrote to Dr. Cassandra Moseley, vice president for research at CSU, and to CSU President Amy Parsons to express scientific and ethical concerns about the studies and request an investigation into the need to kill thousands of animals. The nonprofit also reached out directly to the primary investigator.

“Dietary studies investigating the effects of pulse-rich diets on the gut microbiome and noncommunicable disease outcomes are ethically and effectively conducted using human volunteers,” Janine McCarthy, acting director of research policy for the Physicians Committee, wrote in the letter to Parsons.

Additionally, she wrote, although a search for alternatives to animals is required by federal regulation and the university’s Institutional Animal Care and Use Committee (IACUC), the search conducted by the researcher was flawed, resulting in the primary investigator’s failure to consider viable alternatives to animal use. The IACUC failed to challenge the faulty search.

In response to a Physicians Committee public records request, the university wrote in an email to Ms. McCarthy on Aug. 11, 2025, that the primary investigator had shut down the experiment on July 15, 2025, and that he did not have any active protocols to conduct animal experiments.

“We are grateful that CSU reconsidered these experiments and decided to stop them,” said Ms. McCarthy. “We hope other universities across the country will follow suit by ending animal experiments and shifting to research approaches that are more accurate, cost-effective, and most importantly, more relevant to humans. We also call on the USDA to end funding for animal experiments for human nutrition and instead invest in modern, human-specific science.”

Note to reporters: To arrange an interview with Ms. McCarthy, please contact Kim Kilbride at 202-717-8665 or kkilbride@pcrm.org.

Founded in 1985, the Physicians Committee for Responsible Medicine is a nonprofit health organization that promotes preventive medicine, conducts clinical research, and encourages higher standards for ethics and effectiveness in research and medical training.

 

 

Danforth Center awards proof-of-concept funding to four scientific teams advancing agricultural innovation





Donald Danforth Plant Science Center




ST. LOUIS, MO., August 18, 2025 - As part of its Startup Initiative, the Donald Danforth Plant Science Center has awarded proof-of-concept (PoC) funding to four scientific teams developing breakthrough technologies aimed at solving global challenges in food security and environmental sustainability.

Established to accelerate the movement of scientific discovery from the lab into the marketplace, the Startup Initiative provides critical early support for promising technologies. A key component of this effort, the PoC fund awards $50,000 grants to internal research teams working on pre-commercial innovations with identified commercial potential.

PoC funds support projects that:

  • Generate additional data to secure intellectual property (IP) and strengthen the Center’s IP portfolio;
  • Enhance the attractiveness of technologies for research collaboration and licensing opportunities; and
  • Advance technologies toward the establishment of new startup companies.

“The PoC fund has proven highly effective at developing marketable technologies,” said Danforth Center President and CEO Jim Carrington, PhD. “Over the past four years, several technologies have been derisked through PoC awards, resulting in three new startup companies.” 

The four newly funded projects include:

Developing Genetic Resistance to Geminivirus Pathogens Plant geminiviruses cause numerous devastating diseases in crops such as cassava, tomato, maize, soybean and cotton. The Taylor Laboratory has identified mutations in DNA polymerase delta subunit 1 (POLD1) that confer geminivirus resistance. This discovery offers an important target for biotechnology to generate genetic resistance to geminiviruses across multiple crops, offering farmers an economical and sustainable solution to protect yields.

Platform for Rapid Discovery of Custom Weed Control Strategies Herbicides are widely used and important for weed control in agriculture but are over-used and suffer from limitations on how they can be applied. The Umen Laboratory is developing an efficient customized screening platform to identify the genetic bases for herbicide sensitivity in any crop species. This technology will enable rapid development of strategies to help farmers efficiently combat weeds while reducing overall herbicide usage.

Exploiting Plasmodesmata to Control Plant Disease Plasmodesmata, pores that cross the cell wall to allow the movement of important biomolecules between cells, can be readily exploited by many pathogens. The Burch-Smith and Czymmek teams are pioneering new methods to manipulate plasmodesmata functions to develop novel pesticides and improved crop varieties with entirely new modes of action.

Improving Maize for No-Till Cover Cropping Management Systems  While no-till cover cropping has been demonstrated to be better for the soil, they have also been associated with yield drag in early years, which disincentivizes their adoption by producers. The Baxter and Topp laboratories have developed an approach to phenotype novel maize germplasm for improved performance in a no-till cover crop system. This will enable the identification of alleles and develop germplasm that allow cash crops to be grown in cover crop fields without sacrificing productivity.

About The Donald Danforth Plant Science Center Founded in 1998, the Donald Danforth Plant Science Center is a nonprofit research institute with a mission to improve the human condition through plant science. Research, education and outreach aim to have an impact at the nexus of food security and the environment and position the St. Louis region as a world center for plant science. The Center’s work is funded through competitive grants from many sources, including the National Science Foundation, National Institutes of Health, U.S. Department of Energy, the Gates Foundation, and through the support of individuals and corporations. 

For additional information or images please contact:  Karla Roeber, kroeber@danforthcenter.org .

 

How AI support can go wrong in safety-critical settings


In study scenario, inaccurate AI linked to bad user decisions



Ohio State University




COLUMBUS, Ohio – When it comes to adopting artificial intelligence in high-stakes settings like hospitals and airplanes, good AI performance and a brief worker training on the technology is not sufficient to ensure systems will run smoothly and patients and passengers will be safe, a new study suggests.

Instead, algorithms and the people who use them in the most safety-critical organizations must be evaluated simultaneously to get an accurate view of AI’s effects on human decision making, researchers say.

The team also contends these evaluations should assess how people respond to good, mediocre and poor technology performance to put the AI-human interaction to a meaningful test – and to expose the level of risk linked to mistakes.

Participants in the study, led by engineering researchers at The Ohio State University, were 450 Ohio State nursing students, mostly undergraduates with varying amounts of clinical training, and 12 licensed nurses. They used AI-assisted technologies in a remote patient-monitoring scenario to determine how likely urgent care would be needed in a range of patient cases.

Results showed that more accurate AI predictions about whether or not a patient was trending toward a medical emergency improved participant performance by between 50% and 60%. But when the algorithm produced an inaccurate prediction, even when accompanied by explanatory data that didn’t support that outcome, human performance collapsed, with an over 100% degradation in proper decision making when the algorithm was the most wrong.

“An AI algorithm can never be perfect. So if you want an AI algorithm that’s ready for safety-critical systems, that means something about the team, about the people and AI together, has to be able to cope with a poor-performing AI algorithm,” said first author Dane Morey, a research scientist in the Department of Integrated Systems Engineering at Ohio State.

“The point is this is not about making really good safety-critical system technology. It’s the joint human-machine capabilities that matter in a safety-critical system.”

Morey completed the study with Mike Rayo, associate professor, and David Woods, faculty emeritus, both in integrated systems engineering at Ohio State. The research was published recently in npj Digital Medicine.

The authors, all members of the Cognitive Systems Engineering Lab directed by Rayo, developed the Joint Activity Testing research program in 2020 to address what they see as a gap in responsible AI deployment in risky environments, especially medical and defense settings.

The team is also refining a set of evidence-based guiding principles for machine design with joint activity in mind that can smooth the AI-human performance evaluation process and, after that, actually improve system outcomes.

According to their preliminary list, a machine first and foremost should convey to people the ways in which it is misaligned to the world, even when it is unaware that it is misaligned to the world.

“Even if a technology does well on those heuristics, it probably still isn’t quite ready,” Rayo said. “We need to do some form of empirical evaluation because those are risk-mitigation steps, and our safety-critical industries deserve at least those two steps of measuring performance of people and AI together and examining a range of challenging cases.”

The Cognitive Systems Engineering Lab has been running studies for five years on real technologies to arrive at best-practice evaluation methods, mostly on projects with 20 to 30 participants. Having 462 participants in this project – especially a target population for AI-infused technologies whose study enrollment was connected to a course-based educational activity – gives the researchers high confidence in their findings and recommendations, Rayo said.

Each participant analyzed a sequence of 10 patient cases under differing experimental conditions: no AI help, an AI percentage prediction of imminent need for emergency care, AI annotations of data relevant to the patient’s condition, and both AI predictions and annotations.

All examples included a data visualization showing demographics, vital signs and lab results intended to help users anticipate changes to or stability in a patient’s status.

Participants were instructed to report their concern for each patient on a scale from 0 to 10. Higher concern for emergency patients and lower concern for non-emergency patients were the indicators deemed to show better performance.

“We found neither the nurses nor the AI algorithm were universally superior to the other in all cases,” the authors wrote. The analysis accounted for differences in participants’ clinical experience.

While the overall results provided evidence that there is a need for this type of evaluation, the researchers said they were surprised that explanations included in some experimental conditions had very little sway in participant concern – instead, the algorithm recommendation, presented in a solid red bar, overruled everything else.

“Whatever effect that those annotations had was roundly overwhelmed by the presence of that indicator that swept everything else away,” Rayo said.

The team considered the study methods, including custom-built technologies representative of health care applications currently in use, as a template for why their recommendations are needed and how industries could put the suggested practices in place.

The coding data for the experimental technologies is publicly available, and Morey, Rayo and Woods further explained their work in an article published at AI-frontiers.org.

“What we’re advocating for is a way to help people better understand the variety of effects that may come about from technologies,” Morey said. “Basically, the goal is not the best AI performance. It’s the best team performance.”

This research was funded by the American Nurses Foundation Reimagining Nursing Initiative.