Retail investors are often encouraged to take charge of their financial futures, but the documents they rely on are notoriously difficult to read. In the United States, regulators require investment funds to provide summary prospectuses to help individuals understand what they are buying. These summaries are intended to be plain-English versions of longer statutory documents.
Despite these rules, the summaries continue to feature technical industry terms like leverage, beta, and liabilities. They can also stretch across multiple pages. For individuals without formal financial training, this terminology acts as a barrier to understanding a fund’s strategy, risks, and fees.
Recent advances in artificial intelligence have introduced a new tool for investors. Large language models can instantly translate dense financial jargon into simple, everyday language. A recent study published in the Journal of Behavioral and Experimental Finance investigated whether using AI to simplify fund prospectuses actually improves an investor’s comprehension and changes their likelihood of investing.
Testing the impact of simplified text
Researchers Zihan Gong and Sebastian Müller of the Technical University of Munich wanted to know if text complexity still matters in an era when individuals can generate their own summaries. They set out to measure how different levels of linguistic complexity influence a retail investor’s ability to process information and make decisions.
They designed two online experiments. The first experiment, conducted in May 2024, involved 305 retail investors. The researchers collected 60 real-world summary prospectuses, splitting them evenly between mutual funds and exchange-traded funds (ETFs). Because the original documents averaged thousands of words, the researchers used the language model GPT-4 to generate two 300-word summaries for each fund.
One version was prompted to be “easy,” specifically written for a general audience with no financial background. The other was prompted to be “hard,” written for experienced retail investors with a strong grasp of financial concepts. Each participant read one easy version and one hard version for different funds. After reading, participants rated the text on readability, understandability, and comprehensiveness. They also indicated their willingness to invest in the fund.
Refining the experiment
The researchers conducted a second experiment in December 2025 with 314 participants. This second test served as the main analysis for the study. The researchers adjusted their text generation process to ensure a more realistic comparison.
In the first experiment, the “hard” versions were deliberately complex, which might have introduced an artificial level of difficulty. For the second experiment, the researchers replaced the hard version with a “shortened” version. This version preserved the professional terminology of the original prospectus but was strictly matched in length and structure to the easy version. This design ensured the researchers were testing the effects of plain language rather than just differences in word count.
To see if existing knowledge influenced the results, the researchers also measured the financial literacy of the participants. They used a ten-question objective test covering topics like inflation, interest rates, and diversification. Participants also completed a subjective self-assessment, rating their own confidence in their financial knowledge.
Clearer text leads to higher investment interest
Across both experiments, participants consistently gave higher ratings to the easy versions of the text. In the main experiment, the easy-to-understand summaries raised the combined text accessibility score by approximately 11 percent compared to the shortened versions featuring technical jargon.
This preference for simple language extended to financial intentions. Participants reading the easy versions were 10 percent more likely to express a willingness to invest in the fund. The data showed that simplifying the text did not just make the documents easier to read, but it actively increased the participants’ stated interest in committing money.
The researchers checked whether a participant’s level of financial literacy changed how they responded to the simpler text. It did not. Both financial novices and highly knowledgeable investors benefited from the plain English summaries. Participants with higher objective and subjective financial literacy scores rated the easy texts as more accessible, just as the less knowledgeable participants did.
There was a difference in how the two types of financial knowledge related to investment intentions. While high self-assessed financial knowledge was linked to a higher willingness to invest, objective test scores had no significant association with investment willingness. A participant’s confidence in their own knowledge played a larger role in their intention to invest than their actual test performance.
Familiarity and limits
The study also revealed a sequence effect. Participants rated whichever text they read second as clearer and more understandable than the first. The researchers attribute this to a contrast or familiarity effect, where participants simply became more comfortable with the structure of the task. However, this boost in perceived clarity on the second reading did not translate into a higher willingness to invest.
The findings point to the utility of language models in stripping jargon out of financial disclosures, potentially encouraging wider participation in investment markets. Simplifying terminology benefits a broad spectrum of retail investors, not just beginners.
The study relies on self-reported intentions rather than tracking real-world financial behavior with actual capital at risk. Because the financial literacy test was administered in an unsupervised online setting, some participants might have consulted outside resources to answer the questions. The experimental design also compared AI-generated summaries against other AI-generated summaries, rather than against the original multi-page prospectuses. The researchers chose this approach to prevent participant fatigue, leaving open the question of how much informational value is lost during the initial summarization process.
