How retail investors use GenAI to navigate stock markets

As more investors use AI, their information search may become linked to market activity

Featured faculty: Wu Fan
Written by Joanne Madrid

Retail investors traditionally rely on financial news, analyst reports, and online forums to interpret the market. With generative artificial intelligence (GenAI) rewriting the industry playbook, investors are rapidly deploying it into their decision-making. Deloitte expects the share of individuals using GenAI for investment advice to reach 78 per cent by 2028.

“GenAI significantly lowers retail investors’ mental workload by integrating complex data and offering personalised information aggregation much faster than traditional methods,” says Wu Fan, Assistant Professor of Accounting at the Chinese University of Hong Kong (CUHK) Business School. “Yet little is known about the dynamics of retail investors’ interactions with GenAI.”

Having partnered with one of China’s largest GenAI platforms to analyse more than 1.7 million stock-related queries, Professor Wu finds that users typically start with simple questions, such as asking for stock recommendations or market conditions.

When GenAI answers convey a positive signal, and users perceive it as useful, the queried stock tends to yield higher returns on the same day. This effect is stronger when users react to those answers, such as giving a thumbs-up or sharing them with their networks.

However, Professor Wu warns, this doesn’t mean GenAI accurately predicts or even causes market movement. “GenAI aggregates and echoes existing market sentiment rather than exerting an independent influence on trading behaviour. When users trust GenAI’s answers, they might also act on them, and this collective interest could push the stock price up.”

Highly active stock-related queries often signal a few things: a large number of shares are being traded, some informed traders have better information about certain stocks than others, and the gap between the highest stock price a buyer is willing to pay and the lowest price a seller is willing to accept is widening.

Most common queries from retail investors

In a paper titled, How stock market participants use generative artificial intelligence: Evidence from user-platform interaction data, Professor Wu, Frank Ecker of the Frankfurt School of Finance and Management, Li Xitong of HEC Paris, and Li Yilan of ESSEC Business School investigate how Chinese retail investors begin their GenAI journeys. The data shows 40 per cent of users ask just one stock-related question, but only 8.5 per cent follow up with more than 20 queries.

For retail investors who ask more questions, their requests gradually move from general to more specific queries, such as financial statement analysis, assessing the impact of news events, and comparing competitors. This pattern is the most common among financially knowledgeable users, who tend to use GenAI for deeper analysis.

“Retail investors’ journey evolves from passive information consumption to a more targeted information extraction,” Professor Wu adds. “Investors initially use AI to provide clear basic information and then transition to deep-dive analytical needs.”

GenAI delivers the strongest results when asked to compare, organise, and interpret specific financial information. “Tasks involving structured reasoning and synthesis, such as financial statement analysis, seem to yield the most informational benefits,” he adds.

“GenAI aggregates and echoes existing market sentiment rather than exerting an independent influence on trading behaviour.”

Professor Wu Fan

Do retail investors ask the right questions?

The study also spots a missed opportunity. Retail investors rarely use GenAI to summarise company filings and disclosures, one task where AI excels and could add value.

“Retail investors may not fully understand what GenAI can do, or they simply experiment with the technology out of curiosity without deeply exploring its capabilities, so they still struggle to formulate effective prompts to get the summary they want,” Professor Wu says. “Some may also worry about the accuracy of GenAI summaries and prefer to rely on traditional sources.”

Retail investors often prefer information that is already summarised by intermediaries rather than having GenAI create it from scratch. As more sophisticated investors tend to focus on analytical queries, summarisation may also fall into basic tasks they quickly move past, or not be seen as the most efficient way to obtain the insights they seek.

What triggers GenAI queries

Retail investors typically don’t go straight to GenAI for analysis, but often get their first inspiration from reading or watching the news. User queries rise around major corporate events, such as earnings announcements and performance forecasts, and particularly surge only after such events make headlines.

“GenAI does not completely replace the information funnel,” Professor Wu says. “Platform query volumes still closely track contemporaneous media coverage, suggesting that users often still rely on traditional channels to initiate their research.”

GenAI queries peak during trading hours, as users are likely looking for quick checks of time-sensitive information. After hours, a significant number of queries are still submitted, but they are more likely to be for in-depth research.

User queries also tend to decrease when companies publish reports covering longer or broader topics. More detailed disclosures and performance forecasts are associated with fewer GenAI queries, suggesting that when companies provide investors with enough context upfront, there is less need to seek it elsewhere.

Users will also stay engaged when earlier queries about market signals align with actual stock performance, indicating that perceived accuracy builds trust and repeat use. Surprisingly, users react negatively or show lower interaction to long or complex answers, but respond positively to concise, opinion-driven, and direct responses.

“If answers from GenAI simply mimic the density of traditional analyst reports, retail investors may disengage,” Professor Wu adds.

Preparing for AI-assisted investing

As GenAI adoption grows, Professor Wu suggests that companies rethink how they communicate with investors and make disclosures easier to process by both investors and AI tools.

“Firms must comply with regulatory requirements to provide sufficient, relevant and timely information to the market, but they can also consider structuring disclosures in more machine-readable formats to better facilitate AI-assisted processing,” he says.

Given that users want quick answers and GenAI platforms are good at extracting summaries, corporate disclosures should be clear, concise, and use a consistent style to make it easier for AI tools to identify and extract relevant information. The report should also highlight key takeaways and, instead of just presenting numbers, include an explanation of what they mean and why they changed, so GenAI can analyse the context to provide more helpful explanations.

In terms of content, the report should directly address common user questions about the business outlook, financial performance, and operations, to help GenAI provide better information and potentially reduce the need for users to ask follow-up questions.

“Keep in mind that, when using GenAI, users must remain cautious regarding hallucinations or factual errors in AI-generated responses.” Ultimately, GenAI can be a powerful research tool, but it works best when paired with critical thinking, not as a replacement for it.