Tag: Faculty

Artificial intelligence tools such as ChatGPT and DALL-E are prompting uncertainty about the future employment landscape. Will they help workers by boosting productivity and enhancing the quality of their output? Or will the improved effectiveness of these tools replace workers instead?

As it turns out, the early economic effects of generative AI are already noticeable, if you know where to look.

Olin researchers Oren Reshef and Xiang Hui (along with a colleague from New York University, Luofeng Zhou) set out to quantify how AI tools have affected specific subsets of the labor market—writing-related and image-related freelance workers.

“There’s a lot of uncertainty trying to predict what the effects of AI will be in 10 years, but AI could be very different in 10 years,” said Reshef, an assistant professor of strategy. “We were thinking, ‘Let’s try to see empirically what the direct effect is in the very short run.’”

In both subsets of workers, the effect was evident: In the months after the release of popular AI tools, freelance workers in affected fields saw decreases in both the number of jobs and total earnings recorded on a widely used freelance platform.

The results of this research are the subject of a working paper co-authored by Hui, Reshef and Zhou, “The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market,” which is currently under review for publication.

‘A more immediate effect’

Reshef and Hui said the team chose freelancers to see the rapid effects of AI releases. Because freelance work is naturally short-term and flexible, employers could pivot that work more quickly to AI than they could with full-time employees.

“There is little friction in this marketplace,” Hui said. “So, when you have technological advance, you would expect to see a more immediate effect.”

The researchers studied this market by collecting data from Upwork, an online freelance marketplace for knowledge workers. They could view workers’ employment histories, skills and qualifications, past earnings and reviews.

They started with workers in writing and related fields (proofreading and copy editing), because research has shown that they are among the most vulnerable to disruption by AI. Specifically, they looked at what happened to their jobs and income on Upwork shortly after the November 2022 release of ChatGPT, a powerful language-processing chatbot that creates higher-quality text than was previously available via AI.

After ChatGPT’s release, the number of monthly jobs for writing-related freelancers on Upwork declined by 2%, while monthly earnings declined by 5.2%.

“AI is really getting better, and this year we have ChatGPT, which is like a bomb going off, making everybody aware of it,” Hui said.

Hui, Reshef and Zhou’s analysis showed that after ChatGPT’s release, the number of monthly jobs for writing-related freelancers on Upwork declined by 2%, while monthly earnings declined by 5.2%.

A similar impact hit image-related workers (designers, image editors and artists) after the releases of two image-based AI tools, DALL-E, in April 2022, and Midjourney, in July 2022. Workers in those fields saw even steeper declines—a 3.7% drop in monthly jobs and a 9.4% loss of income.

While the team expected this outcome, they said it was important to be able to show it empirically.

“This is concrete evidence based on real economic data, instead of intuition or hunches,” Hui said.

Documenting the AI transition

Reshef and Hui were struck by the fact that these negative effects didn’t diminish for freelancers who were identified as more experienced and skilled.

“One hypothesis is that the highest-quality people, those with the most experience, would be the ones most protected from this technology—that their skills would act as a moderating factor,” Reshef said.

Instead, the evidence suggested that more highly skilled workers—as defined by such attributes as the number and skill level of past jobs, hourly rate, and Upwork’s “top-rated” badges—actually saw a greater drop in new jobs and income.

While it seems counterintuitive, he said this outcome is consistent with other research on the effects of AI in the workforce.

“In a [separate] experiment, researchers introduce the technology to various workers,” he said. “They find that it helps the lower-quality workers the most, and, if you think about it, it makes sense. If you’re doing lower-quality work, this new tool can help increase your productivity or quality. For higher-quality workers, instead of being protected, you’re losing your competitive edge.”

Reshef and Hui cautioned that the results regarding skill level are not definitive, but rather are suggestive of that outcome.

They also noted that all these results only reflect the short-term impact of AI tools on freelance workers. As AI technology evolves and becomes more integrated into other tools, these effects might be increased or reduced, or provide benefits to workers or consumers to mitigate them.

“We only look at the short-run effect on these workers,” Reshef said. “We’re showing one potential ‘danger’ of this technology, but there are numerous benefits as well that are beyond the scope of this research.

“People could get a new job working with AI to do something we haven’t even thought of yet.”

With divorces, remarriages, living together and childbearing outside of marriage, American families have experienced massive changes in the past 30 years.

The changes are hitting their kids in the wallet, to the advantage of some and disadvantage of others. More than one-third of parents with wills plan to divide their estates unequally among their children, according to the paper “Unequal Bequests,” newly published in the European Economic Review.

The research also discovered that a staggering number of older Americans don’t have a will. Thirty percent of people 70 and older don’t.

Plans for unequal inheritances are concentrated in “weak relationships.” Those are defined as families with stepchildren and families with genetic children with little or no contact with their parents.

Stepparents and parents with no contact with their kids may be less motivated than parents in traditional families to provide resources to children with whom they don’t share genes or haven’t shared homes, said Robert Pollak, Hernreich Distinguished Professor of Economics, and his coauthors.

Children in weak relationships also may be less willing than children in traditional families to assist disabled older parents, especially those with whom they have no genetic connection or only briefly shared a home, according to the paper.

Children of cohabiting couples who break up and repartner are substantially more likely than other children to have weak relationships with adults. 

‘Much more unequal’

“We have shown that bequests are much more unequal now than in the recent past and much more unequal than generally recognized,” Pollak said.

In simple families (families without stepchildren or no-contact children), equal bequests are the dominant pattern. In other families, however, the research found substantial inequality in bequest intentions and final dispositions.

The research used a large and nationally representative US sample from the Health and Retirement Study (HRS) from 1995–2014. The HRS is a survey of more than 24,000 Americans over 50 who are interviewed every two years. If a respondent has a spouse or partner, the spouse or partner is invited to become an HRS respondent.

Economic theory on end-of-life transfers assumes that individuals, or at least older ones, make wills, Pollak said.

“We find, instead, that parents often fail to write wills and, either by design or by default, rely on intestacy law to determine the distribution of their estates.”

When people die without a valid will, their property passes by “intestate succession” to heirs according to state law. In other words, if you don’t have a will, the state will make one for you.

“One important question, which we cannot answer in this paper, is why so many individuals do not write wills.”

Robert Pollak

For parents with stepchildren, relying on intestacy law means leaving everything to genetic and legally-adopted children and nothing to stepchildren. For no-contact parents, the effect is to treat contact and no-contact genetic children equally.

‘Puzzling’

“If parents understand the most basic provisions of the intestacy default, this finding is puzzling,” the authors write.

“It implies that parents who have had no contact with some of their genetic children are more likely to treat all of their genetic children equally than are parents who have maintained contact with all of their genetic children.”

The probability of having a will increases with age. But of the HRS respondents who died between 1995 and 2012, nearly 40 percent died intestate.

Weak relationships

A distinguishing feature of the research is its focus on “weak relationships.” Think of parents with stepchildren and parents with genetic children with whom they have limited or no contact.

The proportion of parents ages 50 and over with more than one child who reported having wills in which they treated their children unequally increased steadily between 1995 and 2014, from around 27% to more than 36%.

Parents with high school or lower educations are about 13% more likely not to have wills than parents with college degrees or higher (47 versus 34%, respectively). The differences by race are larger: Three-quarters of Black and Hispanic parents don’t have a will. About one-third of their white non-Hispanic counterparts don’t.

Standard economic models ignore weak relationships, Pollak said. The assumption is that all children are born to a married couple who remain married to each other. When one spouse dies, the surviving spouse is assumed not to remarry.

“By ignoring divorce, re-partnering or remarriage, economic models fail to recognize the increased complexity of the family,” he said.

Marco Francesconi, of the University of Essex, in the UK, and Domenico Tabasso, of the World Bank-UNHCR Joint Data Center on Forced Displacement in Denmark, were Pollak’s coauthors.

Since 2016, Jian Cai, a senior lecturer in finance at WashU Olin, has advised student practicum teams through the Wells Fargo Advisors Center for Finance and Accounting Research (WFA-CFAR). In a discussion with Tatiana Vdovina, a CFAR PhD Scholar and PhD candidate in finance, Cai talked about the challenges and rewards of guiding students through real-world consulting projects:

What types of companies have you advised in quantitative finance and fintech?

When the practicum program was still in an early stage, two teams consulted with Build-a-Bear (Workshop). In the first year, the company wanted to figure out how to improve their supplies purchasing process. The next year, a team of seven worked on a post-audit analysis of their newly opened stores and presented it to the CFO and an executive committee. Both teams were invited to their flagship store at the Galleria Mall and built our own bears!

Our next client was Detalus, a local investment company. The main project we did with them was to establish a red flag system that would indicate when a company they invested in was predicted to have a credit rating downgrade.

We completed four projects with Neocova, a local startup fintech firm that provides AI services to community banks. (Among those projects were) a predictive model for community banks’ credit rating and an anti-money laundering (AML) model.

Most recently, my team worked with MARKIT/S&P Global Market Intelligence to evaluate the market potential for loan-level probability of default and loss given default modeling. The team won the Quantitative Finance Practicum Showcase because of the sophistication of our analysis and modeling.

What are the biggest challenges you face with a team of master’s students during a 14-week practicum?

For all projects, the first challenge is to translate a business situation or issue the client wants to address into a tangible research question. Most students have not had any industry experience by the time they work on the project. With a lot of motivation and strong technical skills, students tend to be eager to start on data analysis and coding before seeing the actual purpose of the project. So I try to pull them back and help them see the big picture first.

While every semester and every project is different, one thing that is always the same is that time goes by very fast and we need to keep track of the timeline and project scope constantly.

What gives you the most satisfaction in advising student practicum teams?

Being able to solve a seemingly impossible problem makes me happy and intellectually inspired!

For the community banking project (for Neocova), I felt stuck not knowing how to show (that) a group with perceived low credit quality actually has strong creditworthiness. One of our faculty members (the late Radha Gopalan) suggested developing a credit rating predictive model using data we had on rated banks, then making projections on mostly unrated community banks. Thanks to his brilliant idea, we were able to solve the problem proposed by the client quickly.

Ultimately, hearing from students years later and seeing their careers take off makes me happy.

Three students in the most recent practicum team I advised (for S&P) found nice jobs within three months after completing the project. One is now at Morningstar in New York, and another is working at Goldman Sachs. Starting their careers with the experience that they got from the project is awesome to observe.

Read Cai’s full interview and learn more about how student practicum projects have translated into hands-on student experience while providing tangible business benefits to clients.

Pictured at top: Jian Cai’s student practicum team, which worked on projects for Build-A-Bear Workshop, celebrated with newly made stuffed toys.

I wanted to share some exciting news about our incoming faculty member, Michael Mazzeo, dean designate of WashU Olin Business School and professor of business economics.

Mike Mazzeo will be appointed as the Knight Family Endowed Professor in recognition of his leadership in research and teaching. He joined Northwestern University’s Kellogg School of Management in 1998 after completing his PhD in economics at Stanford University.

Most of Professor Mazzeo’s research includes empirical industrial organization, developing new statistical methodologies for examining the relationship between product differentiation strategies and market competition. His work spans a variety of industries, including airlines, banking, healthcare, lodging, retail and telecommunications.

Leading researcher

Professor Mazzeo’s research has been published in leading journals, including The RAND Journal of Economics, the Journal of Economics & Management Strategy, Marketing Science, The Review of Economics and Statistics, the Review of Industrial Organization, Quantitative Marketing and Economics and The Journal of Industrial Economics. Mike is also a renowned teacher. He is a three-time recipient of the Chairs’ Core Course Teaching Award at Kellogg. He also won three best elective teaching awards in Kellogg’s MBA and executive MBA programs.

The Knight Family Endowed Professorship was established this year, and Mike will be its first holder. Joanne Knight and her late husband, Charles F. Knight, have also established the Charles F. and Joanne Knight Distinguished Professorship in Neurology and the Charles F. and Joanne Knight Distinguished Professorship in Orthopaedic Surgery.

Longtime benefactors

Joanne, Charles, and their son, Lester, have been longtime benefactors of Washington University in St. Louis. The Knight family’s history of giving has left an indelible mark on the university. The Knights have been pivotal in establishing WashU Olin as one of the premier institutions of business education and research in the world and ensuring its continued success among top business schools. The family has also contributed significantly to the School of Medicine, where much of their support, and Joanne’s ongoing generosity, has been focused on Alzheimer’s disease research.

Joanne was a charter member of the Siteman Cancer Center Community Advisory Board and served on Washington University’s School of Art National Council. She received an honorary doctor of humanities degree from the university in 2010 and a Second Century Award from the School of Medicine in 2019. Joanne and Charles received the Dean’s Medal from Olin in 2012.

Please join me in congratulating Professor Mike Mazzeo. We will have a formal installation ceremony at a later date.

Philip Dybvig, the Boatmen’s Bancshares Professor of Banking and Finance at Olin Business School at Washington University in St. Louis and co-recipient of the 2022 Nobel Prize in economics, was a key speaker at the 2023 Asian Leadership Conference, Korea’s premier international conference, in May.

The annual conference provides a platform for global leaders to gather and discuss the most pressing issues facing the world today. This year’s theme was “The era of upheaval: The road to collaboration and innovation.” Other key speakers included Mario Draghi, former prime minister of Italy; Boris Johnson, former prime minister of the United Kingdom; Robert Barro, a professor of economics at Harvard University; and Oliver Stone, Academy Award-winning director.

Dybvig’s session — moderated by Sung Taeyoon, a professor of economics at Yonsei University in Korea — delved into the phenomenon of the financial crisis, including the bankruptcy of major financial institutions occurring worldwide and the reality of bank runs. He also discussed the potential impact of U.S. monetary policy and liquidity issues in the financial market on the crisis and various policy proposals and countermeasures to help stabilize the financial market.

Sung (left) and Dybvig speak at the 2023 Asian Leadership Conference. (Image courtesy of ALC)

WashU Olin Professor Salih Tutun and his team took the Gold Award (first place) in the May 2023 IISE Cup Competition. Their research led to an innovation that can quickly identify common mental disorders, such as depression and anxiety.

The team designed a way to map with color imaging 10 common mental disorders. Mental health professionals can use the product to screen patients and monitor the patients’ progress during treatment, and it’s already in use at a clinic in Turkey.

“Mental health experts need innovative tools to efficiently screen a large volume of potential patients for mental health disorders,” said Tutun.

The world is facing an unprecedented mental health crisis. An estimated 970 million people, nearly one in eight people globally, lived with mental disorders in 2019, according to the World Health Organization. The crisis has only gotten worse since the COVID-19 pandemic. In the United States, one in five adults experience mental disorders today, National Alliance on Mental Illness reports.

According to a Lancet Commission report, the overall cost of mental health problems between 2016 and 2030 would surpass $16 trillion globally due to lost productivity, disability, social welfare spending for homelessness and poverty, and law-and-order spending for public safety.

“The demand for mental health services is exploding around the globe,” Tutun said. The need is growing for automated tools to support screening, diagnosis and monitoring, he said.

MDscan

The working paper about the research is titled “MDscan: An explainable AI artifact for screening and monitoring mental disorders.”

Tutun and his team presented their findings May 21 at the IISE competition in New Orleans. The Institute of Industrial and Systems Engineers runs the annual international competition to showcase solutions to business or social problems. Tutun’s team beat out competition from Amazon, Airbus, IBM, DHL and others.

The study used computational design science to develop the explainable artificial intelligence (XAI) artifact—named MDScan—for screening, diagnosing and monitoring disorders. Explainable artificial intelligence is a set of processes and methods that allows human users to comprehend and trust the results and output that machine learning algorithms create.

MDscan uses patient responses to a clinical neuropsychological test to create full-color images, similar to radiological images. It predicts which disorder or combination of disorders is afflicting a patient and the severity. It also explains the underlying logic of the prediction.

The researchers evaluated MDscan’s performance by using patient data collected from 200 mental health clinics. MDscan outperformed current clinical practice with an average of 20% improvement, according to the research.

MDscan helps doctors manage their patient load, Tutun said, by delivering a preliminary diagnosis and classifying patients by mental condition and severity.

Shortage of mental health providers

While the demand for mental health services grows, many places suffer from a shortage of providers. Some countries report as little as one psychiatrist for every 100,000 people. The result: long wait times for appointments, delayed diagnosis and treatment, and continued suffering. Psychiatrists and counselors have called for innovative tools to boost their capacity to screen and treat patients more efficiently and effectively, Tutun said.

But clinics haven’t broadly accepted other AI tools, he said, because of their “black box” models, which lack transparency. “Without explanations, experts cannot trust these predictions for clinical use, where patients’ lives and well-being are often at stake,” Tutun said.

MDscan is a “white box” model that elaborates on what patient answers contributed to the predicted diagnosis and to what extent.

Tutun is an expert in decision analysis, deep learning, design science and explainable AI for understanding how human behaviors and social interactions affect critical decisions in business.

In addition to Tutun, team members were Anol Bhattacherjee of the University of South Florida, Kazim Topuz of the University of Tulsa, Ali Tosyali of the Rochester Institute of Technology, and Gorden Li of Bosch Center for Artificial Intelligence.

“The MDscan artifact presented in this paper attempts to address this critical problem while also demonstrating how to improve explainability, user trust, confidence, and acceptance of AI predictions by explaining their AI predictions,” the authors wrote.

“We hope that our research will inspire other researchers to develop their own XAI approaches for addressing critical business and social problems in other domains, such as finance and cybersecurity, where visual and explainable representations of large complex data sets are needed to detect essential and useful patterns.”