Vendors
日本語
search
  2. //
  3. Reports & insights
  4. //
  5. Meaningful integration of Generative AI in Capital Markets set to happen Sooner than Later

Meaningful integration of Generative AI in Capital Markets set to happen Sooner than Later

Create a vendor selection project
Click to express your interest in this report
Indication of coverage against your requirements
A subscription is required to activate this feature. Contact us for more info.
Celent have reviewed this profile and believe it to be accurate.
We are waiting for the vendor to publish their solution profile. Contact us or request the RFX.
Projects allow you to export Registered Vendor details and survey responses for analysis outside of Marsh CND. Please refer to the Marsh CND User Guide for detailed instructions.
Download Registered Vendor Survey responses as PDF
Contact vendor directly with specific questions (ie. pricing, capacity, etc)
19 July 2023
Patrick Wegner

Generative AI continues to be in the headlines in recent weeks, mostly recently with the launch of Llama V2, which is open source and includes free commercial use. The excitement around Generative AI remains strong, but in conversations with market participants, we have more balanced discussions, with awareness around challenges increasing.In the last few weeks, there have been reports in the press regarding lawsuits and inquiries against LLM providers, such as an author’s lawsuit in the UK and the Federal Trade Commission (FTC) probe into ChatGPT in the US. These developments come on top of major data sources such as Reddit considering charging for using its data for AI training purposes.

Like other sectors, capital markets are trying to figure out how they will use and interact with the new technology. This new technology poses plenty of risks in such a heavily regulated industry. Data security and ownership, explainability, bias, hallucinations/inaccuracies, and output ownership are top of mind and concerns echoed by a recent roundtable Celent hosted on the subject. These risks are likely why several prominent capital markets participants have been reported to have limited or banned the use of LLMs in their companies.

Just Monday, SEC chairman Gary Gensler gave us some pointers to areas regulators might focus on regarding AI. In his speech, he mentioned the areas cited above. However, he also noted concerns about use cases focusing on “rent extraction” rather than fulfilling the firms’ fiduciary duty, which requires many financial services actors to put the needs of their clients before theirs. Gensler added that SEC Staff has been asked to “make recommendations on rule proposals for the Commission’s consideration.”

On a more macro scale, there are concerns about herd mentality decreasing financial stability, especially if applications are fundamentally built off a small number of foundational models. An oligopoly is potentially the case, given the advantages of scale and limited access to the vast amounts of data necessary to build the models. The CFTC also discussed the topic at a Technical Advisory Committee (TAC) meeting on Tuesday, which included a series of presentations on “Responsible AI in Regulated Financial Services” as the CFTC continues to monitor the space.

That said, all this is entirely normal when it comes to the evolution of new technologies or workflows. This evolution is true for AI as it is for ESG or Blockchain. First, come the niche adopters, then the hype, followed by a framework-building, standards creation, and regulatory review process before a new technology is fully adopted across financial markets use cases. If anything, it feels like Generative AI is going through this process faster than other new technologies. This shortened time frame is possibly due to the transformative abilities of Gen AI and the competitive advantage this promises or some of the groundwork previously laid by other forms of AI. Or perhaps the time is right, with the cloud, a key enabler for AI, data distribution, and advanced data analytics, being more widely adopted across capital markets.

We believe we will see meaningful integration of LLM models into the capital markets framework sooner rather than later. Many of the same companies that have limited the use of Generative AI for their workforce are also actively searching for use cases. I am currently researching use cases for generative AI across capital markets, so keep watch for that report, or in the meantime, please reach out if you would like to discuss what you are seeing.

Comments