Investors increasingly use GenAI to research companies and evaluate stocks

Generative artificial intelligence is quietly transforming the way ordinary people engage with the stock market, and a new study from researchers at Indiana University’s Kelley School of Business, the University of Washington’s Foster School of Business, and the Gies College of Business at the University of Illinois provides the most detailed picture yet of how retail investors actually use these tools. Drawing on an analysis of more than 410,000 investor queries submitted to a major brokerage’s generative AI chatbot, alongside a survey of nearly 2,200 investors, the research finds that almost half of retail investors have already turned to GenAI to process financial information or inform their investment decisions. The study, titled “Generative AI and Investor Processing of Financial Information,” is forthcoming in the Journal of Accounting and Economics, one of the most influential journals in the accounting and finance research community.

The lead author, Joe Croom, assistant professor of accounting at the Kelley School of Business, explains that the dominant use case is not what many might expect. Investors are not primarily treating chatbots as oracles that dispense stock tips. Instead, they are deploying generative AI as a research assistant, a tool for making sense of the enormous and often impenetrable volumes of financial data that companies, regulators, and markets produce every day. Rather than asking simplistic questions such as “What stock should I buy,” most users pose nuanced, analytical queries like “How healthy are this company’s profit margins?” The technology’s central value, according to the study, lies in compressing the time and cognitive effort required to digest complex financial disclosures, earnings reports, and market-moving news into accessible, interpretable summaries.

One of the most striking findings concerns how investor behavior evolves over time as users gain experience with the tools. Croom and his co-authors, Elizabeth Blankespoor, professor of accounting at the University of Washington’s Foster School of Business, and Stephanie Grant, associate professor of accountancy at the Gies College of Business, observed a clear trajectory in the chatbot interaction data. Investors typically begin by using generative AI to screen stocks and perform high-level evaluations of companies, essentially using it as a discovery mechanism to narrow a vast universe of investment options. But as their familiarity with the tool deepens, their usage patterns shift toward more detailed monitoring and interpretation of specific firm-level news, suggesting that GenAI becomes embedded in an ongoing, iterative research workflow rather than serving as a one-time shortcut.

The adoption data reveal a technology in the midst of rapid diffusion. While many users in the dataset appeared to experiment with the chatbot only briefly before abandoning it, a substantial subset, 17.7 percent of users, adopted it as a routine part of their investing process. The survey results reinforce this picture of accelerating uptake: among investors who have used GenAI, 74 percent said they believe it improves their processing of financial information, and 80 percent planned to continue using it in the future. Croom notes that the data were collected in 2024, which means the figures likely understate current usage levels considerably. Given the pace at which large language models have improved and proliferated since then, the true adoption rates today are almost certainly higher and still climbing.

When asked why they embrace the technology, investors point overwhelmingly to cognitive efficiency. Sixty-five percent of those surveyed cited GenAI’s capacity to speed up the processing of information, while 59 percent highlighted its ability to simplify the handling of complex data. These figures speak to a genuine bottleneck in retail investing. Corporate filings, regulatory disclosures, macroeconomic reports, and earnings transcripts present a formidable analytical challenge for individuals who lack the training and time of professional analysts. Generative AI, by translating dense financial language into plain explanations and by synthesizing information across multiple sources, effectively lowers the barrier to sophisticated analysis that was previously accessible only to institutions with teams of analysts and expensive data terminals.

Yet the study is equally notable for documenting the skepticism and concerns that temper this enthusiasm. Investors identified several significant limitations of generative AI for investing tasks. Reliability and accuracy topped the list, with 54 percent of respondents expressing concern about whether the tools produce trustworthy outputs, a worry that resonates with well-documented problems such as hallucinated facts and fabricated citations in large language models. Data privacy followed closely, cited by 50 percent of investors, reflecting unease about sharing sensitive financial information with third-party AI systems. Response quality was a concern for 46 percent. These reservations suggest that even enthusiastic adopters approach the technology with a degree of caution, treating its outputs as provisional rather than definitive.

The divide between users and non-users offers perhaps the most provocative insight into the technology’s future trajectory. Among investors who have not yet used GenAI, more than half, 55 percent, expressed uncertainty about whether it would improve their information processing. However, only 24 percent said they were unlikely to adopt the technology at all. Croom interprets this gap as a signal of future growth: most non-users are neither enthusiastic nor firmly opposed, but merely undecided, and as the tools become more capable and more socially normalized, adoption within this group is likely to expand substantially.

But the research complicates one of the most popular narratives surrounding artificial intelligence and finance: the idea that these tools will democratize markets and level the playing field between novice and expert investors. The data point in the opposite direction. It is the investors with the most financial experience and sophistication who are driving adoption and extracting the most value from generative AI. “Many people expect GenAI to democratize markets and empower less sophisticated investors. We find the opposite: the investors with the most financial experience and sophistication are the ones driving adoption and using GenAI most effectively,” Croom said. This finding carries important implications for market fairness, because if advanced users can process information faster and more deeply than before, the informational advantage of sophisticated investors over casual ones may actually widen rather than shrink. The technology amplifies existing skill rather than substituting for it, a pattern consistent with broader research on how complementary technologies tend to reward those already equipped to exploit them.

The implications extend well beyond individual portfolios. Croom emphasizes that the evidence of widespread retail adoption has direct consequences for regulators, who must now design investor education programs and safeguards for a population increasingly reliant on AI-generated financial interpretations. It also matters for corporate managers, who craft the disclosures that GenAI-using investors will feed into chatbots and whose messaging may be interpreted, summarized, or distorted by algorithms rather than read directly by humans. If a growing share of the investing public encounters a company’s financial story primarily through the lens of an AI summary, the incentives around how firms communicate, and how regulators require them to communicate, may need to be rethought from first principles.

The research was conducted using data provided by Public, an investment brokerage and financial technology company with millions of active users, which gave the researchers rare visibility into real-world query behavior at scale, complementing the self-reported survey evidence. Together, the two data sources allow the authors to move beyond speculation about how people might use AI and toward empirical documentation of how they actually do. As Croom concludes, “As more investors turn to GenAI, we need to keep learning how these tools reshape the way investors process and act on financial information, and what that means for capital markets.” The study marks an early but consequential step in that research agenda, and its central message, that generative AI is already a mainstream tool of retail investing, wielded most powerfully by those who need it least, is likely to reverberate through discussions of market regulation, investor protection, and disclosure policy for years to come.

News Publication Date: 2-Sep-2026

Web References: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5053905

References: Croom, J., Blankespoor, E., & Grant, S. Generative AI and investor processing of financial information. Journal of Accounting and Economics. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5053905