Exame: Generative AI Sparks Enthusiasm but Still has Limitations in the Financial Sector, Says Bloomberg Executive

This story originally ran in Brazil’s Exame in Portuguese.

In an exclusive interview with Exame, Bloomberg’s Global Head of Core Product shares the tech giant’s perspective on artificial intelligence.

The rise of generative artificial intelligence has generated significant excitement across various industries, but it is already clear that the technology has several limitations that hinder its adoption in the financial sector.  That’s the assessment of Wayne Barlow, Global Head of Bloomberg’s Core Product team, in an exclusive interview with Exame.

The executive states that “despite the enthusiasm generated by other technology providers in various industries, GenAI has limitations in the financial domain, and we are highly attuned to this.” The tech and data giant’s strategy in finance revolves around “mitigating” these limitations.

“Generative AI models can hallucinate, quickly become outdated, and, on their own, cannot reason or perform basic calculations. Some of these issues have a significant impact on the financial sector, where a lot is at stake. There is no doubt that AI security concerns will drive regulatory controls in global markets,” he explains.

AI in the Financial Sector

Despite the recent hype surrounding generative AI, Barlow points out that artificial intelligence has been around and used in financial markets for many years. Bloomberg, for instance, began leveraging AI more than 15 years ago. However, projects like ChatGPT, powered by Large Language Models (LLMs), have introduced new dynamics to the sector.

“Generative AI is important for investment professionals because they need new user experiences and intuitive ways to interact with data and analytics. These professionals are also constantly balancing between ‘information overload’ and the ‘fear of missing something important,’” he highlights.

In this context, the executive sees advantages in adopting AI to streamline and optimize news consumption. However, for this to be effective, AI-powered products must be “practical and specifically designed for investment professionals.”

Recently, Bloomberg launched AI-powered news summaries for investors, where generative AI creates the summaries, which are then reviewed by experts to refine them. Barlow notes that “the solution is simple but saves investors time,” which helps drive adoption.

“Investors need help processing and organizing an increasing volume of information to make faster and more informed business and investment decisions,” he emphasizes.

For him, AI must support workflow development that “connects data and unstructured content, such as long documents, across three main areas: helping process data in new ways, generating insights, and transforming how users interact with our products.”

Barlow believes that research analysts focused on equity and credit segments, as well as portfolio managers, are likely to benefit the most from this new technology. Generative AI “pushes boundaries by enabling broad discoveries, deep explorations, and multi-functional synthesis of information.”

Automation and Finance

The advancement of technologies such as artificial intelligence has also driven automation initiatives in the financial markets. At the same time, discussions about its benefits and potential challenges have increased. However, the Bloomberg executive believes that, at this stage, full automation is not feasible.

“Certain tasks will always require human intervention for critical investment decisions. This was already proven during the last wave of information technology when markets became more electronic over the past 20 years,” he points out.

Even so, he believes generative AI will be particularly useful for automating four key areas: data analysis and visualization, insight gathering and synthesis, content creation and publication, and communication management and analysis.

Another area expected to gain momentum in 2025 involves AI agents, which can perform various functions without human supervision or interaction.

Barlow comments that “it’s fascinating to think about thousands of agents and superintelligences while the industry debates whether multiple AI agents, each with expertise in different domains working together to solve complex problems, will gain traction in 2025.”

Regarding Bloomberg’s approach, he shares that “as we evaluate AI agents in the financial sector, we’ve learned some key lessons. Sometimes, they do not communicate the way a human would with their underlying agents. When AI agents tackle a user query, the underlying system must be trained in relevant AI frameworks. Additionally, not all underlying agents are fully based on generative AI.” For this reason, the field still faces several uncertainties before it can scale further.

###