Stressed for a bit? Then don’t click it, cybersecurity experts advise

Phishing psychology study explores what makes workers vulnerable

Peer-Reviewed Publication

DOE/PACIFIC NORTHWEST NATIONAL LABORATORY

 

IMAGE: WORKERS WHO ARE FEELING DISTRESSED ARE MORE LIKELY TO CLICK ON A PHISHING EMAIL, ACCORDING TO NEW HUMAN FACTORS RESEARCH AT PACIFIC NORTHWEST NATIONAL LABORATORY. view more 

CREDIT: ILLUSTRATION BY TIMOTHY HOLLAND | PACIFIC NORTHWEST NATIONAL LABORATORY

RICHLAND, Wash.—Workers feeling a specific form of stress are more likely than others to become the victims of a phishing attack, according to a study at the Department of Energy’s Pacific Northwest National Laboratory.

 

While most—if not all—of us feel stress in the workplace, scientists identified a specific form of stress that indicates who is more vulnerable to clicking on bogus content that could lead to malware and other cyber ills. The work could help workers and their employers increase their cybersecurity defenses by recognizing the warning signs when someone is about to make a risky click.

 

The team’s results from a study of 153 participants were published recently in the Journal of Information Warfare. The researchers noted that while the relatively small sample size limited their ability to tease out all of the relationships among more than two dozen variables they studied, the relationship between stress and response to the simulated phishing email was statistically significant.

 

The costs of phishing attacks are enormous. An analysis sponsored by Proofpoint and conducted by the Ponemon Institute estimates that large U.S. businesses lost, on average, $14.8 million apiece to fraudsters via phishing in 2021 alone.

 

Defenses include not just better technology but also improved awareness by would-be victims.

 

“The first step to defend ourselves is understanding the complex constellation of variables that make a person susceptible to phishing,” says PNNL psychologist Corey Fallon, a corresponding author of the study. “We need to tease out those factors that make people more or less likely to click on a dubious message.”

 

In their study, Fallon and colleagues found that people who reported a high level of work-related distress were significantly more likely to follow a phony phishing email’s link. Every one-point increase in self-reported distress increased the likelihood of responding to the simulated phishing email by 15 percent.

 

The scientists describe distress as a feeling of tension when someone on the job feels they’re in a difficult situation and unable to tackle the task at hand. Distress might stem from feeling their workload is too high, or they might be questioning whether they have adequate training or time to accomplish their work.

 

 

Fancy phish to explore phishing psychology

 

The 153 participants had agreed to take part in a study, but they were unaware that the phishing email sent a few weeks later was part of the planned study into human factors research.

 

As far as phishes go, this was a fancy phish. There was no mention of a large sum of money from an African prince, for example, and there were no outright spelling mistakes or gross grammatical errors.

 

“These were well-crafted emails deliberately designed to trick people and tailored to the organization,” said Jessica Baweja, a psychologist and an author of the study. “It was much harder to detect than the average phish.”

 

Each participant received one of four different versions of a message about an alleged new dress code to be implemented at their organization. The team tested three common phishing tactics separately and together. Here’s what they found:

• Urgency. 49 percent of recipients clicked on the links. Sample text: “This policy will go into effect 3 days from the receipt of this notice...acknowledge the changes immediately.”
• Threat. 47 percent clicked. “…comply with this change in dress code or you may be subject to disciplinary action.”
• Authority. 38 percent clicked. “Per the Office of General Counsel…”
• The three tactics together: 31 percent clicked.

 

While the team had expected that more tactics used together would result in more people clicking on the message, that wasn’t the case.

 

“It’s possible that the more tactics that were used, the more obvious it was a phishing message,” said author Dustin Arendt, a data scientist. “The tactics must be compelling, but there’s a middle ground. If too many tactics are used, it may be obvious that you’re being manipulated.”

 

In day-to-day operations, PNNL tests its staff with fake phishing emails periodically. Typically around just 1 percent of recipients will click. Far more employees spot the phish early on and provide crowd-sourced alerting to the Laboratory’s cybersecurity experts, said Joseph Higbee, PNNL’s chief information security officer. When a real phishing email is detected, the Laboratory purges the system of all instances of the email immediately. The information is frequently shared with other DOE laboratories.

 

Human-machine teaming to reduce cybersecurity risk

How can companies and employees use this data to reduce the risk?

 

“One option is to help people recognize when they are feeling distressed,” said Fallon, “so they can be extra aware and cautious when they’re especially vulnerable.”

 

In the future, one option might be human-machine teaming. If an algorithm notes a change in a work pattern that might indicate fatigue or inattention, a smart machine assistant could suggest a break from email. Automated alerts are becoming more common, for instance, when a driver drifts unexpectedly and the car issues a warning about fatigue. The researchers noted that the potential benefits of input from a machine assistant would need to be weighed against employee privacy concerns.

 

“It can be hard to see email as a threat,” said Baweja. “Our ancient brains aren’t wired to equate email with scary things. You’re working through emails all day and it’s routine; there’s little reason to think they could harm you or our organization.

 

“Organizations need to be thinking about how to encourage people to make good choices. People overestimate their ability to detect phishing emails,” she added.

 

PNNL researchers are continuing the work, but with a twist. Instead of asking what makes people more vulnerable to phishing, they will conduct a small study of people who resisted the bait, to learn more about their traits and state of mind as they monitor their email.

 

The work is part of a broader program in human-machine teaming and human factors research at PNNL, which recently hosted a Symposium on Human Factors.

 

The work was funded by the Cybersecurity and Infrastructure Security Agency, part of the Department of Homeland Security. In addition to Arendt, Baweja and Fallon, authors include Ji Young Yun and Nick Thompson of PNNL and Zhuanyi Shaw, formerly of PNNL.

 

# # #

---

METHOD OF RESEARCH

Observational study

SUBJECT OF RESEARCH

People

ARTICLE TITLE

Phishing in the Wild: An Ecologically Valid Study of the Phishing Tactics and Human Factors that Predict Susceptibility to a Phishing Attack

Medical cannabis laws have negligible impact on opioid prescribing

Peer-Reviewed Publication

WEILL CORNELL MEDICINE

State laws allowing medical cannabis use did not reduce prescriptions for opioids or other therapies for chronic, non-cancer pain, according to a policy analysis by Weill Cornell Medicine investigators.

Dr. Beth McGinty, chief of the Division of Health Policy and Economics at Weill Cornell Medicine, used a study design that emulates a clinical trial to analyze the effects of medical cannabis laws on treatment for people with chronic non-cancer pain. The results, published July 4 in Annals of Internal Medicine, challenge the results of previous studies that suggested medical cannabis laws might reduce opioid prescribing, helping curb the ongoing opioid crisis in the United States.

“Some research suggests that perhaps medical cannabis laws reduce opioid prescribing for chronic non-cancer pain because some people may substitute cannabis,” said Dr. McGinty, who was recruited as a professor of population health sciences at Weill Cornell Medicine. “We found no effects of these laws on opioid prescribing or any types of treatment for chronic non-cancer pain that we looked at.”

Public health authorities and policymakers hoping to stem the ongoing opioid overdose crisis in the United States have taken various approaches, including recommending more judicious opioid prescribing. Some have suggested allowing medical cannabis to give patients with chronic non-cancer pain an option with a lower risk of overdose. But many questions remain about this approach.

“The science is very much still evolving on how effective cannabis is for chronic non-cancer pain,” Dr. McGinty said. Questions remain about how effective cannabis is at treating chronic non cancer pain compared with evidence-backed therapies, including non-opioid pain medications and physical therapy.

To understand how medical cannabis laws affect chronic non-cancer pain treatment, Dr. McGinty and her colleagues used commercial insurance claims data to analyze opioid prescribing, non-opioid pain medication prescribing, and delivery of chronic non-cancer pain procedures like physical therapy in 12 states with medical cannabis laws and 17 states without such laws. The team used statistical adjustments to correct for any pre-law differences between pain treatment outcomes in medical cannabis states and their comparison states.  

“It’s an observational study, but we set it up in a way that mimics a clinical trial as closely as possible,” she said.

The study did not find a significant impact on the number of patients with chronic non-cancer pain receiving any prescription opioid, any non-opioid prescription pain medication, or on procedures used to relieve chronic pain. Dr. McGinty said the results were consistent across states with medical cannabis laws.

“Medical cannabis laws do not appear to be associated with changes in the prescription opioid or other non-cannabis, non-opioid treatments for chronic non-cancer pain,” she said.

Dr. McGinty acknowledged that the study has limited statistical power, in part because there are only 50 states, a limitation in any study of the effects of state policies. Investigators could also examine the effects of medical cannabis laws on chronic pain treatment using a data source that captures medical cannabis use in addition to data on other chronic pain treatments, for example state prescription drug monitoring programs that are in some cases starting to include medical cannabis as well as opioid prescriptions.  

In the meantime, the study may provide valuable information for policymakers looking to understand how medical cannabis laws influence prescribing of prescription opioids and delivery of other treatments for chronic non-cancer pain.

“Our study findings suggest that medical cannabis laws do not significantly reduce opioid prescribing,” Dr. McGinty said. “Policy makers trying to curb excess opioid prescribing and overdoses should focus on other strategies.”

---

JOURNAL

Annals of Internal Medicine

ARTICLE TITLE

Effects of U.S. State Medical Cannabis Laws on Treatment of Chronic Noncancer Pain

ARTICLE PUBLICATION DATE

4-Jul-2023

HOSPITALS KILL

Kenyan hospital visits linked to increased exposure to antibiotic-resistant bacteria

Peer-Reviewed Publication

WASHINGTON STATE UNIVERSITY

PULLMAN, Wash. -- Kenyan patients who spend more than three days in the nation’s hospitals are more likely to harbor a form of bacteria resistant to one of the most widely used antibiotic classes, according to a recent study led by Washington State University.

The research team found that 66% of hospitalized patients were colonized with bacteria resistant to third-generation cephalosporins, compared to 49% among community residents. Third-generation cephalosporins are typically used for serious infections, and resistance to these antibiotics leaves limited options for treating patients with some bacterial infections.

The study, published in the Clinical Infectious Diseases journal, was part of a pair of projects in Kenya and a third in Guatemala to determine the prevalence of antibiotic-resistant bacteria. The research also aimed to identify risk factors for colonization with bacteria resistant to important and frequently used classes of antibiotics, including third-generation cephalosporins.

“These bacteria can cause untreatable infections,” said Sylvia Omulo, an assistant professor at WSU’s Paul G. Allen School for Global Health who led the studies in Kenya. “We have to use surveillance to make sure we understand what causes these bacteria to colonize and, later, resist certain antibiotics.”

By examining health records of Kenyan hospital patients who were colonized with cephalosporin-resistant bacteria, researchers identified three risk factors associated with colonization. Those risk factors included hospitalization for more than three days at 132% increased risk, intubation at 73% and a positive HIV status at 70%.

A person colonized with the bacteria may show no symptoms of an infection; however, the germs are present and growing in their body, and the individual may be at increased risk of a more dangerous infection from something as simple as a minor medical procedure or sickness from common bacteria like E. coli and Klebsiella. The germs can unknowingly be spread through person-to-person contact or contaminated surfaces.

While resistance is often associated with improper antibiotic use, that does not appear to be a primary contributing factor for cephalosporin-resistant bacteria in either Kenya or Guatemala. 

The Guatemala study, led by WSU Allen School assistant professor Brooke Ramay, was interrupted by the COVID-19 pandemic and resulted in data being collected during two time periods. Reported antibiotic use in the community declined three-fold between the pre- and post-periods, but the prevalence of cephalosporin-resistant bacteria remained steady, with colonization found in 67% of hospital patients and 46% of community subjects.

“If antibiotic use locally in the community was a major risk factor, we'd expect some change over the course of the year,” said WSU Regents Professor Douglas Call, corresponding author for the three publications.

 

In a Kenyan study of communities, the likelihood of colonization with cephalosporin-resistant bacteria rose by 12% with increasing visits to hospitals and clinics, and individuals who kept poultry were 57% more likely to have the resistant bacteria. Previous and ongoing work in both countries also highlights the role of bacterial transmission due to poor sanitation and hygiene.

What is not clear is if contact with the health care system is a source of transmission or if people seeking care are more likely to harbor these bacteria.

“We know if you're interacting with the health care system, you're more likely to carry these bugs, but we don't know why yet,” Call said. “To sort out the cause from the effect, we need to track the same people over time and record how their colonization status changes with different behaviors. Studies to do this are being set up for the coming year.”

The research was completed in collaboration with the Centers for Disease Control and Prevention, University del Valle de Guatemala, University of Nairobi and the Kenya Medical Research Institute. The work was funded by the CDC and is part of the broader Antimicrobial Resistance in Communities and Hospitals (ARCH) study partnership, which has research projects in six countries. The most recent findings from WSU and other institutions involved in the ARCH study were published in a special supplement of the Clinical Infectious Diseases journal.

---

JOURNAL

Clinical Infectious Diseases

DOI

10.1093/cid/ciad258 

ARTICLE PUBLICATION DATE

5-Jul-2023

Factors associated with learning disabilities and autism led to requests for euthanasia and assisted suicide in the Netherlands, study finds

A Kingston University, London study found reasons for chosing euthanasia and assisted suicide included feelings of loneliness and a struggle to form friendships

Peer-Reviewed Publication

KINGSTON UNIVERSITY

A Kingston University, London study found several people with learning disabilities and autism in the Netherlands chose to die legally through euthanasia and assisted suicide due to feeling unable to cope with the world, changes around them or because they struggled to form friendships.

There were almost 60,000 cases of euthanasia and physician-assisted suicide between 2012 and 2021 in the Netherlands. More than 900 case reports were recorded on an online, searchable database. This form of dying is legal in the country provided six due care criteria are met, one of which is that the patient’s suffering is unbearable, with no prospect of improvement.

Study lead and Professor of Intellectual Disability and Palliative Care at Kingston University Irene Tuffrey-Wijne and colleagues examined the database to understand the situations where people with learning disabilities or autism spectrum disorders had been granted requests for euthanasia or assisted suicide, and the main causes of suffering that led to the request.

They found 39 cases that involved people with either learning disabilities or autism spectrum disorder, or in some cases, both. Of these, 19 were male and 20 were female, with 18 of the 39 under the age of 50.

Published in the BJPsych Open Journal, the study found a majority, around two-thirds, chose that path solely, or mostly, due to characteristics associated with their condition, such as anxiety, loneliness and not feeling they had a place in society.  In eight of the 39 cases the reasons for the individual’s suffering were exclusively related to the learning disability or autism, with people feeling unable to cope with the world, changes around them and a struggle to make friendships and connections.

The research team said the findings highlighted a need for an international philosophical and ethical debate around criteria for euthanasia and assisted suicide and how to deal with such requests from vulnerable patient groups.

Within the study, a further eight said their autism or learning disability made it hard to cope with non-life-threatening physical illness or decline, while the same number stated their condition combined with psychiatric problems like anxiety and loneliness was their reason for wanting to die. Of the 39 cases, causes of suffering that led people to asked for help to die included loneliness (77 per cent of cases), a dependence on others (62 per cent), a lack of resilience or coping strategies (56 per cent) and a lack of flexibility and difficulty adapting to change (44 per cent).

Finally, in a third of the cases, doctors noted explicitly that the individual’s learning disabilities or autism were not treatable and that this was a key consideration in their assessment that there was no prospect of improvement and therefore an assisted death was the only remaining option for the patient.

Commenting on the findings, Professor Tuffrey-Wijne said that alongside the need for a wider debate on the ethics of dealing with requests from vulnerable groups, the research also highlighted the importance of investing in highly individualised support services around the world. “What these individuals needed was a society where they felt they belonged, with people around them who not just accept and accommodate, but welcome their unique ways of being,” she said.

“There is no doubt some people feel helpless, suffer deeply and profoundly and the support needed is complex but hopefully there can be a culture shift and a big investment in services that are significantly under-resourced.”

In addition to this work, Professor Tuffrey-Wijne gave oral evidence at a government health and social care select committee about an assisted dying and suicide inquiry earlier this month.

---

JOURNAL

BJPsych Open

DOI

10.1192/bjo.2023.69 

METHOD OF RESEARCH

Data/statistical analysis

SUBJECT OF RESEARCH

People

ARTICLE TITLE

Euthanasia and physician-assisted suicide in people with intellectual disabilities and/or autism spectrum disorders: investigation of 39 Dutch case reports (2012–2021)

AI unleashes a Pandora's box: ChatGPT generates convincingly fake scientific article

Researchers demonstrate how AI can generate seemingly authentic scientific articles, prompting ethical concerns in the scientific community

Peer-Reviewed Publication

JMIR PUBLICATIONS

 

IMAGE: AI-GENERATED IMAGE, IN RESPONSE TO THE REQUEST "PANDORAS BOX OPENED WITH A PHYSICIAN STANDING NEXT TO IT. OIL PAINTING HENRY MATISSE STYLE", (GENERATOR: DALL-E2/OPENAI, MARCH 9, 2023, REQUESTOR: MARTIN MÁJOVSKÝ). view more 

CREDIT: CREATED WITH DALL-E2, AN AI SYSTEM BY OPENAI; COPYRIGHT: N/A (AI-GENERATED IMAGE);

A new study published in the Journal of Medical Internet Research on May 31, 2023, by Dr Martin Májovský and colleagues has revealed that artificial intelligence (AI) language models such as ChatGPT (Chat Generative Pre-trained Transformer) can generate fraudulent scientific articles that appear remarkably authentic. This discovery raises critical concerns about the integrity of scientific research and the trustworthiness of published papers.

Researchers from Charles University, Czech Republic, aimed to investigate the capabilities of current AI language models in creating high-quality fraudulent medical articles. The team used the popular AI chatbot ChatGPT, which runs on the GPT-3 language model developed by OpenAI, to generate a completely fabricated scientific article in the field of neurosurgery. Questions and prompts were refined as ChatGPT generated responses, allowing the quality of the output to be iteratively improved.

The results of this proof-of-concept study were striking—the AI language model successfully produced a fraudulent article that closely resembled a genuine scientific paper in terms of word usage, sentence structure, and overall composition. The article included standard sections such as an abstract, introduction, methods, results, and discussion, as well as tables and other data. Surprisingly, the entire process of article creation took just 1 hour without any special training of the human user.

While the AI-generated article appeared sophisticated and flawless, upon closer examination expert readers were able to identify semantic inaccuracies and errors particularly in the references—some references were incorrect, while others were non-existent. This underscores the need for increased vigilance and enhanced detection methods to combat the potential misuse of AI in scientific research.

This study’s findings emphasize the importance of developing ethical guidelines and best practices for the use of AI language models in genuine scientific writing and research. Models like ChatGPT have the potential to enhance the efficiency and accuracy of document creation, result analysis, and language editing. By using these tools with care and responsibility, researchers can harness their power while minimizing the risk of misuse or abuse.

In a commentary on Dr Májovský’s article, published here, Dr Pedro Ballester discusses the need to prioritize the reproducibility and visibility of scientific works, as they serve as essential safeguards against the flourishing of fraudulent research.

As AI continues to advance, it becomes crucial for the scientific community to verify the accuracy and authenticity of content generated by these tools and to implement mechanisms for detecting and preventing fraud and misconduct. While both articles agree that there needs to be a better way to verify the accuracy and authenticity of AI-generated content, how this could be achieved is less clear. “We should at least declare the extent to which AI has assisted the writing and analysis of a paper,” suggests Dr Ballester as a starting point. Another possible solution proposed by Majovsky and colleagues is making the submission of data sets mandatory.

The article “Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora's Box Has Been Opened” was published by JMIR Publications in its flagship title, the Journal of Medical Internet Research and can be accessed here.

Please cite as:

Májovský M, Černý M, Kasal M, Komarc M, Netuka D Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora’s Box Has Been Opened J Med Internet Res 2023;25:e46924 

URL: https://www.jmir.org/2023/1/e46924/

doi: 10.2196/46924

 

###

 

About JMIR Publications

JMIR Publications is a leading, born-digital, open access publisher of 30+ academic journals and other innovative scientific communication products that focus on the intersection of health, and technology. Its flagship journal, the Journal of Medical Internet Research, is the leading digital health journal globally in content breadth and visibility, and is the largest journal in the medical informatics field.

To learn more about JMIR Publications, please visit jmirpublications.com or connect with us via Twitter, LinkedIn, YouTube, Facebook, and Instagram.

Head office: 130 Queens Quay East, Unit 1100, Toronto, ON, M5A 0P6 Canada

Media contact: communications@jmir.org

The content of this communication is licensed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, published by JMIR Publications, is properly cited.

---

JOURNAL

Journal of Medical Internet Research

DOI

10.2196/46924 

ARTICLE TITLE

Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora’s Box Has Been Opened

AI tests into top 1% for original creative thinking

Reports and Proceedings

THE UNIVERSITY OF MONTANA

 

IMAGE: ERIK GUZIK IS AN ASSISTANT CLINICAL PROFESSOR IN THE COLLEGE OF BUSINESS AT THE UNIVERSITY OF MONTANA. view more 

CREDIT: UM PHOTO BY RYAN BRENNECKE

MISSOULA – New research from the University of Montana and its partners suggests artificial intelligence can match the top 1% of human thinkers on a standard test for creativity.

The study was directed by Dr. Erik Guzik, an assistant clinical professor in UM’s College of Business. He and his partners used the Torrance Tests of Creative Thinking, a well-known tool used for decades to assess human creativity.

The researchers submitted eight responses generated by ChatGPT, the application powered by the GPT-4 artificial intelligence engine. They also submitted answers from a control group of 24 UM students taking Guzik’s entrepreneurship and personal finance classes. These scores were compared with 2,700 college students nationally who took the TTCT in 2016. All submissions were scored by Scholastic Testing Service, which didn’t know AI was involved.

The results placed ChatGPT in elite company for creativity. The AI application was in the top percentile for fluency – the ability to generate a large volume of ideas – and for originality – the ability to come up with new ideas. The AI slipped a bit – to the 97th percentile – for flexibility, the ability to generate different types and categories of ideas.

“For ChatGPT and GPT-4, we showed for the first time that it performs in the top 1% for originality,” Guzik said. “That was new.”

He was gratified to note that some of his UM students also performed in the top 1%. However, ChatGTP outperformed the vast majority of college students nationally.

Guzik tested the AI and his students during spring semester. He was assisted in the work by Christian Gilde of UM Western and Christian Byrge of Vilnius University. The researchers presented their work in May at the Southern Oregon University Creativity Conference.

“We were very careful at the conference to not interpret the data very much,” Guzik said. “We just presented the results. But we shared strong evidence that AI seems to be developing creative ability on par with or even exceeding human ability.”

Guzik said he asked ChatGPT what it would indicate if it performed well on the TTCT. The AI gave a strong answer, which they shared at the conference:

“ChatGPT told us we may not fully understand human creativity, which I believe is correct,” he said. “It also suggested we may need more sophisticated assessment tools that can differentiate between human and AI-generated ideas.”

He said the TTCT is protected proprietary material, so ChatGPT couldn’t “cheat” by accessing information about the test on the internet or in a public database.

Guzik has long been interested in creativity. As a seventh grader growing up in the small town of Palmer, Massachusetts, he was in a program for talented-and-gifted students. That experience introduced him to the Future Problem Solving process developed by Ellis Paul Torrance, the pioneering psychologist who also created the TTCT. Guzik said he fell in love with brainstorming at that time and how it taps into human imagination, and he remains active with the Future Problem Solving organization – even meeting his wife at one of its conferences.

Guzik and his team decided to test the creativity of ChatGPT after playing around with it during the past year.

“We had all been exploring with ChatGPT, and we noticed it had been doing some interesting things that we didn’t expect,” he said. “Some of the responses were novel and surprising. That’s when we decided to put it to the test to see how creative it really is.”

Guzik said the TTCT test uses prompts that mimic real-life creative tasks. For instance, can you think of new uses for a product or improve this product?

“Let’s say it’s a basketball,” he said. “Think of as many uses of a basketball as you can. You can shoot it in a hoop and use it in a display. If you force yourself to think of new uses, maybe you cut it up and use it as a planter. Or with a brick you can build things, or it can be used as a paperweight. But maybe you grind it up and reform it into something completely new.”

Guzik had some expectation that ChatGPT would be good at creating a lot of ideas (fluency), because that’s what generative AI does. And it excelled at responding to the prompt with many ideas that were relevant, useful and valuable in the eyes of the evaluators.

He was more surprised at how well it did generating original ideas, which is a hallmark of human imagination. The test evaluators are given lists of common responses for a prompt – ones that are almost expected to be submitted. However, the AI landed in the top percentile for coming up with fresh responses.

“At the conference, we learned of previous research on GPT-3 that was done a year ago,” Guzik said. “At that time, ChatGPT did not score as well as humans on tasks that involved original thinking. Now with the more advanced GPT-4, it’s in the top 1% of all human responses.”

With AI advances speeding up, he expects it to become a key tool for the world of business going forward and a significant new driver of regional and national innovation.

“For me, creativity is about doing things differently,” Guzik said. “One of the definitions of entrepreneurship I love is that to be an entrepreneur is to think differently. So AI may help us apply the world of creative thinking to business and the process of innovation, and that’s just fascinating to me.”

He said the UM College of Business is open to teaching about AI and incorporating it into coursework.

“I think we know the future is going to include AI in some fashion,” Guzik said. “We have to be careful about how it’s used and consider needed rules and regulations. But businesses already are using it for many creative tasks. In terms of entrepreneurship and regional innovation, this is a game changer.”

###

Researchers find hidden micro-stressors in routine driving

University of Houston and Texas A&M Transportation Institute use affective computing to gain unique insights into the driver condition

Peer-Reviewed Publication

UNIVERSITY OF HOUSTON

 

IMAGE: IOANNIS PAVLIDIS, ECKHARD-PFEIFFER DISTINGUISHED PROFESSOR OF COMPUTER SCIENCE AT THE UNIVERSITY OF HOUSTON view more 

CREDIT: UNIVERSITY OF HOUSTON

Commuters around the world dream of ideal driving conditions every day, but they rarely get them. The traffic is often heavy and gets worse when the weather turns sour. Light traffic and good weather are usually perceived as factors for a stress-free commute. Alas, researchers from the University of Houston and the Texas A&M Transportation Institute found that even under such ideal conditions, daily driving is stressful to many people, and for intriguing reasons.

In a study published in the journal IEEE Transactions on Affective Computing, researchers report that people with a predisposition to anxiety exhibit a significantly higher heart rate when they drive as opposed to those who are not predisposed to anxiety. Anxious drivers had a heart rate about five beats per minute higher than non-anxious drivers, under similar conditions.

The study’s researchers also found drivers’ heart rate increased significantly with car speed. Drivers moving at 65 mph had a heart rate of about four beats per minute higher than drivers moving at 25 mph, under similar conditions. In both cases, the observed cardiovascular activation was linked to sympathetic activation, that is, to stress responses.

“These are substantial numbers that we would have never guessed”, said Ioannis Pavlidis, Eckhard-Pfeiffer Distinguished Professor of Computer Science, who led the University of Houston group. “Anxious people who commute at highway speeds experience, on average, heart rate elevation of nine beats per minute; this is every day for an hour or more, which is the typical commuting time in this country. It is an unexpected stressor that is hard to ignore because of its substantial effect and its repetitive nature.”

Both Mike Manser, senior research scientist and Robert Wunderlich, director of the Center for Transportation Safety with the Texas A&M group, emphasized that the reported stress effect was measured under good weather and light traffic, that is, ideal commuting conditions. “This sizeable stress effect stands to grow even bigger, as the conditions get worse”, they said.

The researchers conducted the driving study with a new affective computing method they call NUBI, short for Naturalistic and UBIquitous. The researchers said they recruited volunteers from Texas, where most people commute daily, the weather is usually good and traffic conditions are ideal in mid-size cities. All the volunteers were young (mean age 27), healthy and experienced drivers.

Monitoring was completely unobtrusive. For a week, every few seconds the drivers’ heart rate was monitored through their own Apple Watches, while their driving data was monitored through OBD II devices, popular with car insurance tracking programs. The volunteers also had to respond to daily psychometric questionnaires they received via text messages when they were not driving. Momentary weather and surrounding traffic data corresponding to the volunteers’ driving and heart rate data were extracted via open internet sources based on the drivers’ GPS signal. The researchers also monitored the volunteers during their non-driving hours to establish the uniqueness of any driving effects.

“Across the state, we were shadowing the participants’ real-lives moment by moment for a week,” Pavlidis said. “This is the beauty of a NUBI driving study – you can monitor participants hundreds of miles away and get insights into their natural driving and non-driving behaviors.”  

The resulting data set was huge, including 77 hours of driving and 915 hours of non-driving observations, expressed as nearly four million rows of multimodal information, which were analyzed with machine learning and statistical models.

The results of this study follow the results of an earlier study by the same researchers, who reported that certain people experience significant stress even in minute acceleration events, such as stop and go traffic. It appears there are cohorts in the population for which even trivial driving is stressful. These daily micro-stressors were not known or suspected thus far, because NUBI-style studies with unprecedented access to drivers’ internal and external state were not feasible until recently.

“Because driving is ingrained into people’s lives, even individuals who exhibit the said stress responses are not consciously aware of them. Nevertheless, the responses are there, they are substantial and their long-term implications are unknown,” said Pavlidis.  

For the short term, these micro-stressors appear to overload the drivers who experience them, because for similar itineraries, afflicted drivers consistently report being more tired than non-afflicted drivers, he added. Collectively, the study’s long- and short-term results have potential lifestyle, safety and insurance implications, the researchers noted.

---

JOURNAL

IEEE Transactions on Affective Computing

DOI

10.1109/TAFFC.2023.3291330 

SUBJECT OF RESEARCH

People

ARTICLE TITLE

Investigating Cardiovascular Activation of Young Adults in Routine Driving

Public support for militarily defending NATO allies

Peer-Reviewed Publication

PNAS NEXUS

 

IMAGE: EFFECT OF TARGET JOINING NATO, OVERALL AND BY SENDER COUNTRY. view more 

CREDIT: TOMZ ET AL.

Voters in North Atlantic Treaty Organization (NATO) member states are far more willing to militarily defend another country if the country joins NATO, versus if the same country does not join NATO, according to a study. To explore the possible consequences of expanding NATO membership, Michael Tomz and colleagues surveyed 14,000 voters in 13 NATO countries. Each survey participant was presented with a hypothetical Russian attack on one of four possible targets: Bosnia, Finland, Georgia, or Sweden—the four countries (other than Ukraine) furthest along in their bids for NATO accession at the time of the study. Whether or not the target had joined NATO prior to the attack was randomly varied. Participants were asked if their own country should defend the target country militarily. NATO membership boosted support for military defense from an average of 45% to 74% across the four targets. For Bosnia and Georgia, support nearly doubled between the non-member and member conditions. The benefits of joining NATO were smaller for Finland and Sweden, because most voters in NATO countries would defend Finland and Sweden even if those countries remained outside the alliance. Finally, willingness to defend new members of NATO depended critically on whether voters perceived the alliance as valuable for their own country. Anti-NATO rhetoric could, therefore, undermine the alliance by making voters less willing to defend alliance members, whereas pro-NATO rhetoric could enhance defense and deterrence. The results have significant implications for geopolitics and current debates about the expansion of NATO, according to the authors.

---

JOURNAL

PNAS Nexus

DOI

10.1093/pnasnexus/pgad206 

ARTICLE TITLE

How NATO Membership Transforms Public Support for War