Newsroom

Dr. Devin Shanthikumar, associate dean of undergraduate programs at the UCI Paul Merage School of Business, is at the forefront of AI-driven financial research. Her current research spans two important areas: examining how AI can be used to track extreme sentiment on social media platforms like Seeking Alpha, and using that to examine the impact of social media on extremeness, and exploring how AI can improve financial analyst forecasting. Taken together, these distinct research projects reveal the profound effect AI will have on financial markets.

Tracking Extremism on Seeking Alpha

Dr. Shanthikumar’s groundbreaking working paper, “Sentiment Extremeness on Social Media: Evidence from Seeking Alpha Comments,” is coauthored with Qiao Annie Wang, Montana State University, and Shijia Wu, The Chinese University of Hong Kong, Shenzhen, both Ph.D. graduates from the UCI Paul Merage School of Business. This research examines how financial social media platforms like Seeking Alpha, a widely used financial platform where investors share their views on stocks and market trends, affect user sentiment and extremism.

“Social media can drive us to become more extreme, but it could also have the opposite effect—helping us learn from each other and meet closer in the middle,” Dr. Shanthikumar says. Her research aims to determine how these phenomena play out in financial contexts​. She and her team analyzed vast comment datasets from Seeking Alpha, using AI to measure whether users’ sentiments were positive or negative and how extreme those sentiments were.

The results were surprising: Participation on Seeking Alpha tended to moderate users’ opinions over time, making them less extreme rather than more. This moderation effect challenges the common assumption that social media inherently amplifies extremism. “When someone comments early on—then they comment again and again—they become more and more moderate as they discuss the stock,” says Dr. Shanthikumar​. This effect is particularly strong when users aren’t anonymous, suggesting platform design and transparency play a crucial role in shaping user behavior.

Their research implications are broad. Platforms like Seeking Alpha, which expose users to a variety of perspectives, can reduce echo chambers and encourage more balanced discussions. This, in turn, leads to a decrease in volatility, as more moderate sentiment tends to result in less erratic trading behavior. “When you have more extreme sentiment, you get more trading volume, more volatility. But as extremeness decreases, that excess trading volume goes away,” Dr. Shanthikumar says.

She stresses how integral the AI language tool was to their research. “One of the key takeaways from our paper is that you can use large language models very well to measure the type of free-form, regular conversational text that most people use. For example, the AI seemed to catch when people were being sarcastic or using words in a nontraditional way. It was very good at capturing true sentiment.” As she ponders ways to teach AI use to her students, this is an important takeaway, she says.

Improving Analyst Forecasts with AI

Dr. Shanthikumar also explores how AI affects the productivity and accuracy of financial analysts’ forecasts. A new working paper, “Artificial Intelligence and Analyst Productivity,” coauthored with Il Sun Yoo, Shidler College of Business, University of Hawai‘i at Mānoa, another UCI Paul Merage School of Business Ph.D. graduate, addresses three key questions:

1. Does AI improve the quantity and quality of forecasts?
2. Does it complement or replace human analysts?
3. How does AI investment affect employment in the financial sector?

Their research reveals that AI significantly increases the quantity of forecasts financial analysts produce. More importantly, the quality of these forecasts also improves, especially for more volatile stocks that are harder to predict. “The AI allows the analysts to do what they do best and to do it better,” she says. “It’s basically complementing the analyst’s skills.”

This is a crucial finding because it demonstrates AI can enhance human performance in complex tasks rather than simply automating them. However, AI’s effect on employment is more nuanced. While analysts’ productivity increases, according to the research, investment banks that employ AI decreased hiring of new analysts. “They didn’t lay anyone off, but they reduced subsequent hiring,” Dr. Shanthikumar says. “It’s important to note that this is the short-term effect on employment. The existing analysts are doing more, and they’re doing it better, so the firm doesn’t need more analysts. But it remains to be seen if firms then figure out new and better ways to use analysts and subsequently return to hiring more.” The vocational implications should be examined further in future research, she believes.

One of the other interesting aspects of the research is how different types of AI are used. Dr. Shanthikumar distinguishes between broad AI systems that analyze complex financial data and more accessible tools like ChatGPT. While ChatGPT can increase the quantity of work, it does not improve the quality of forecasts as broadly. “ChatGPT might help us write emails and reports, but it will not necessarily help with complex numerical analysis for forecasting earnings,” she says. This highlights the need for specialized AI systems tailored to specific financial tasks.

Ramifications for Investors and the Financial Industry

Dr. Shanthikumar’s AI research offers valuable insights for investors and the financial industry. On social media platforms like Seeking Alpha, AI reveals that certain key factors influence the moderation of sentiment to create a more balanced and informed investing environment. Investors should be aware that platforms designed to expose users to diverse opinions, rather than reinforce existing beliefs, can provide a more accurate picture of market sentiment and reduce the risk of making decisions based on extreme or emotionally charged views. In terms of AI, the research shows that AI can be used to analyze investor sentiment in a systematic way.

With regard to financial analysis, Dr. Shanthikumar’s work shows AI is not only a tool for automation but a powerful complement to human expertise. Investors can take comfort in the fact that AI-backed forecasts are likely to be of higher quality, especially for uncertain or volatile firms. This suggests banks and financial institutions with significant AI investments may offer more reliable forecasts, which is something investors should pay attention to.

For financial institutions, Dr. Shanthikumar’s findings underscore the importance of thoughtful AI integration. While AI can improve productivity and forecast quality, firms need to be mindful of its effect on employment and ensure it supplements, rather than replaces, human analysts. “Our research shows that AI is currently most useful to complement what humans do best, not to replace us. Let’s not try to automate knowledge-based jobs,” she says. “Instead, let’s think about how we can take advantage of AI to let people do what they do well.”

The Future of AI in Finance

As AI continues to evolve, its role in both financial analysis and investor behavior will grow. Dr. Shanthikumar’s research is the beginning of what promises to be a transformative period for the financial industry. In the future, she plans to explore how AI affects other aspects of analysts’ work, such as their written reports and participation in conference calls​.

Moreover, the socioeconomic implications of AI adoption in finance cannot be ignored. While AI-driven productivity gains are beneficial, the potential for reduced hiring in knowledge-based industries raises important questions about AI’s effect on the future of white-collar work in finance and beyond. “We’re seeing a little bit less new hiring, even in highly educated fields. That does raise questions going forward,” she says.

AI’s Implications: A Balanced Approach

Dr. Shanthikumar’s dual approach to AI in financial research—exploring both its effect on analyst productivity and its role in assessing user sentiment online—offers a comprehensive view of how AI is reshaping the financial landscape.

Her findings suggest that while AI can enhance human performance, careful consideration must be given to its broader implications, from employment to market behavior.

As AI technology continues to advance, its integration into financial markets will require a balanced approach, one that leverages its strengths while safeguarding the human element that remains critical to interpreting and navigating the complexities of financial forecasting.