Investors have always faced a massive challenge: processing an absurd volume of financial information every single day without missing a detail that could make or break a buy or sell decision.
When the Wall Street Journal started testing AI-generated summaries on its articles back in 2024, the most common reaction was skepticism.
The concern was fair enough: what if artificial intelligence oversimplified the news, leaving out exactly the context that matters most for people operating in the market?
But a new study out of Harvard Business School came in with data that flipped that argument on its head. 📊
Rather than hurting comprehension, AI-generated summaries did exactly the opposite: they pulled readers deeper into the stories, working like a trail of breadcrumbs leading them to the most complex points in each piece. They improved overall understanding of the articles and even sped up market reactions.
Can you call it a surprise?
Maybe not so much when you look at the numbers up close. 👇
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
What the Harvard study actually found
The research was led by professor Joseph Pacelli from Harvard Business School, in collaboration with Tony Cho and Allen Huang from the Hong Kong University of Science and Technology, along with professor Kristina Rennekamp from Cornell University. Together, they authored the paper titled Generative AI and Investor Processing of Financial Media, which analyzed the behavior of real Wall Street Journal readers during the period when AI summaries were rolled out across the publication’s articles.
To reach their findings, the researchers examined 1,734 articles containing financial analysis of 158 companies, published between July 2024 and June 2025. The goal was to understand whether those automatically generated text blocks had any measurable impact on engagement, comprehension, and most importantly, the decisions of investors who follow financial news on a daily basis.
One interesting detail is the lead researcher’s own initial stance. Pacelli, who holds the Gerald Schuster Associate Professor of Business Administration chair, admits he started the study as a skeptic. In his words, the expectation was that AI-generated summaries might not be all that useful, potentially leading to quick and shallow readings. What the data showed, however, was the opposite: these summaries help hold attention and inform the reader, which is great for markets.
Faster market reactions
The study uncovered something pretty compelling for anyone who follows the financial market closely. By looking at each article’s effect on a company’s stock within the 30 minutes following online publication, the researchers found that companies mentioned in stories with AI summaries saw a 3.5% increase in trading volume.
This suggests that AI not only helped people better understand what they were reading but also accelerated the decision-making process. In the world of investors, that speed can be the difference between entering a position at the right moment or missing a window of opportunity entirely. Stock prices also adjusted more quickly when articles included summaries, which suggests, according to Pacelli, that traders are getting informed faster and creating more efficient markets.
And there’s more: the observed effects were even stronger when articles covered complex topics or when there was heavy competition from overlapping news and earnings releases happening at the same time. In other words, the summaries help the most during moments of overload, exactly when readers need a hand getting through a heavy pile of content.
Like an appetizer that works up your appetite
To better understand how readers process these summaries, Pacelli and his colleagues set up a lab experiment. They gathered 124 graduate students and handed them two identical Wall Street Journal articles, one with an AI summary and one without, then tested how much each group could remember about the content.
The numbers speak for themselves:
- Those who read with the summary remembered more. These readers answered correctly, on average, 4.72 out of 7 questions about the overall content and 2.49 out of 4 questions about information that appeared only in the body of the article, meaning outside the summary.
- Those who read without the summary remembered less. Performance dropped to 3.95 out of 7 questions overall and 2.05 out of 4 on details that were not in the summary.
Notice the most important point here: readers with the summary performed better even on questions about sections that were not in the summary. This dismantles the idea that AI makes people skip the rest of the text. In practice, the summary worked as an entry point that encouraged deeper consumption.
Another data point reinforces this reading. Google searches for information about the companies increased when articles included AI summaries, signaling that people went looking for more data from other sources. As Pacelli himself puts it, the summaries don’t lead to less engagement — if anything, the opposite happens, they seem to generate more engagement. The comparison he makes is a great one: it’s like an appetizer that makes you hungry enough to stay for the full meal. 🍽️
Why investors benefit so much from AI summaries
Anyone who works in market analysis knows the routine all too well: dozens of reports, breaking news, company press releases, macroeconomic data, and the constant need to cross-reference all of it with what’s already in the portfolio. The ability to process that volume of information quickly and accurately is, quite literally, a competitive advantage. This is where AI-generated summaries step in with a very concrete proposition: reducing the time needed to extract the core of a piece of information without sacrificing the quality of understanding.
What artificial intelligence does in these cases isn’t just condensing paragraphs. Modern language models are trained to identify the most relevant points in a text, structure information hierarchically, and present content in a way that makes it easier to absorb, especially in technical, dense contexts like financial journalism. For an investor who needs to make quick decisions based on market news, this kind of assistance represents a real shift in how information gets consumed.
There’s one more point worth highlighting: the democratization of access to quality information. Not every investor has the technical background to interpret a dense Wall Street Journal article with the same ease as a senior analyst. AI summaries create an accessibility layer that levels the playing field, allowing people with different levels of market experience to extract value from content that might have previously felt impenetrable.
An important word of caution
Despite the encouraging results, Pacelli makes sure to pump the brakes on one point. You have to consider the context and the type of content AI is summarizing. In the case of financial journalism, readers are highly motivated to seek information — after all, it’s their money on the line with the companies they’re analyzing. That incentive is what drives them to dig deeper after reading the summary.
But it’s not clear whether the same behavior would repeat with news where the reader has no direct stake. In those cases, the risk of more superficial reading is real. As Pacelli himself notes, it takes a lot more than three bullet points to truly understand a topic like inflation. If this trend keeps pushing forward without careful thought, there’s a danger of readers losing the ability to appreciate the richness of complex issues.
How to use AI summaries well
Thinking about what companies and media outlets can take away from the study, Pacelli offers some very practical recommendations:
- Consider using generative AI to summarize information you share with the public. The logic is simple: everyone is already using their own AI to summarize content. If your company doesn’t do it, the risk is that someone else will do it for you, and likely in a way that reflects their own interests and biases.
- Evaluate using summaries for complicated materials. AI summaries prove most valuable precisely when audiences face information overload or particularly complex content.
- Put humans in charge of reviewing the summaries. Since generative AI can increase the risk of misinformation through biases or hallucinations, these texts demand rigorous fact-checking and careful editing.
What changes from here
With the Harvard Business School study results in hand, it’s likely that more financial publications will start exploring similar models for using AI to generate summaries. The precedent set by the Wall Street Journal is no small thing: this is one of the most influential publications in the world when it comes to finance, and the positive data on comprehension and investor engagement will naturally catch the attention of competitors and other verticals within specialized journalism.
From a technology standpoint, the case also reinforces something the artificial intelligence community has been discussing for a while: the value of large language models isn’t just in generating original content but in the ability to process, organize, and present existing information in ways that expand human understanding. For the financial market, where the speed and accuracy of information have a direct impact on results, this capability has very concrete practical value.
The most relevant takeaway from all of this might be the simplest one: when applied well, AI doesn’t impoverish information. It opens doors. And for investors operating in a market that never stops, having a tool that improves comprehension and accelerates decisions without compromising content quality is exactly the kind of edge that makes a difference day in and day out. 🚀
