GenAI on Wall Street - Opportunities and Risk Controls
Jackie Jianhong ShenFinancial IndustryNew York City, New York, USA, Dedicated to Gil Strang on the Occasion of His 90 Birthday · 2025
Read GenAI on Wall Street - Opportunities and Risk Controls by Jackie Jianhong ShenFinancial IndustryNew York City, New York, USA, Dedicated to Gil Strang on the Occasion of His 90 Birthday (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-05841.txt.
Introduction
Generative AI inside investment banks is surveyed in a 2025 overview by Jackie Shen. Banks are adopting these tools quickly for research, drafting, and client work, chasing speed and new services. Each new use, Shen argues, opens a matching set of risks that firms must control from the start.
The overview walks through emerging applications across the financial industry, with investment banks at the center. Then it turns to the risk side: controls for errors, misuse, and exposure that older systems never posed. Framing the two sides as a yin and yang to be held together, Shen argues firms can let adoption grow organically while guarding the wider system. The piece serves as a map of opportunity paired with the safeguards it demands.
Catalog record
| CREATOR | Jackie Jianhong ShenFinancial IndustryNew York City, New York, USA, Dedicated to Gil Strang on the Occasion of His 90 Birthday |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-05841
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
GenAI on Wall Street - Opportunities and Risk Controls Thanks: Permanent Email & URL: [email protected] & https://alum.mit.edu/www/jhshen. Jackie Jianhong ShenFinancial IndustryNew York City, New York, USA, Dedicated to Gil Strang on the Occasion of His 90 Birthday Abstract We give an overview on the emerging applications of GenAI in the financial industry, especially within investment banks. Inherent to these exciting opportunities is a new realm of risks that must be managed properly. By heeding both the Yin and Yang sides of GenAI, we can accelerate its organic growth while safeguarding the entire financial industry during this nascent era of AI. Keywords: Generative AI (GenAI), Large Language Models (LLM), Agents, Structured, Unstructured, JSON, Coding-Free, Agentic Network, Data Privacy, Information Barriers, Hallucination, Brainwash, Gossips, Generalization. Attention: The current work is designed to be published exclusively on the Social Sciences Research Network ( SSRN ) and the arXiv preprint servers. Any commercial or open-journal publication is strictly prohibited without the prior consent from the author. 1 Introduction 1.1 To Write or Not To Write With the awakening of AI, writing an article like the current one has become a challenge. For example, Google Gemini or Microsoft Copilot can now compose a reasonable version quite swiftly, and virtually on any subject. The challenge is to genuinely convince ourselves, the Human Intelligence (HI), that we can write something truly special, more exquisite, better predictive, and hence more rewarding to the audience than AI could offer. Such confidence starts to drain supersonically when we face reality, especially when AI has become the dearest darling on the red carpet of capitalism. It is heart-wrenching to acknowledge even the slightest hint that AI has been gradually winning over HI, and to forfeit the very pride of HI that has been meticulously sowed since the Stone Ages by our caveman ancestors. After wrestling with our own intelligence this way back and forth, however, we do come to rediscover a precious jewel almost completely lost in this social-media era. That is, “Aha, at least we are able to harvest joy through writing!” To develop a paper or article is a conscious effort of organizing and browsing through one’s experiences, linking disconnected observations, playing with various thought experiments, and finally arriving at a coherent piece of creation. Once composed with sufficient devotion, such an article is no different from a decent pop song or a piece of classical music, when it comes to the joy and satisfaction that reward the inner world of the author(s). An AI agent can also respond with a text string like “Aha, I am very proud of my answer.” But only God knows whether it truly tastes the sweetness of that “pride,” or it is merely an algorithmic parrot that is able to replicate an emotional phrase. Once this unique article is written and published on any archive site, it will be instantly fed into the incessantly crawling AI and absorbed into AI’s next response. In this sense, it is hopeless that HI will ever be able to retain our superiority over AI. But for now, we should at least happily pacify ourselves - “Thank you, Mr. AI. The joy is still mine, though.” 1.2 The Yin and Yang of GenAI Unlike AI answering, HI writing is always based on the limited and often specialized personal exposure to a phenomenon or a field of concern. Therefore, we should first seriously warn the readers that biases, flaws or limitations are inevitable in this writing. Most content here results from the author’s limited working experience in both TradFi (i.e., traditional finance ) and DeFi (i.e., decentralized finance - mainly on cryptos ), as well as via various meetings, conferences, and thematic dialogues with colleagues of similar interests, in both industry and academia. We will cover several emerging applications of generative AI (GenAI) in the financial industry, and mainly focus on the tremendous potential of large language models (LLM) in modern corporate and office environments. These applications may signify a dawning era of AI, as well as a profound shake-up to the inertia of conventional working styles since the very beginning of the industrial revolutions. Numerous use cases or thought experiments will be constructed and discussed in Section 2 . From internal project management, trading, sales to more general client services, all such emerging applications open new doors to unprecedented efficiency and productivity in modern corporations. Meanwhile, the universal entanglement of Yin and Yang (i.e., risks and rewards) commands us to also stay vigilant for the associated risks and controls. This constitutes the second main theme of the current work. In Section 3 , we navigate through several categories of common risks and also propose their remediation measures. Finally, the current work is not specifically devoted to AI-based automated coding or software development. That is a critical area impacting all the engineering and technology teams worldwide, and it deserves a separate in-depth article. We do cover a couple of related examples in which specialized code blocks crafted by GenAI can dramatically improve day-to-day office work. 2 Opportunities of LLM-Based GenAI In this section, we explore several major use cases where LLM-based GenAI could profoundly impact the financial industry. Some of them are universal in the office working environment, though. 2.1 Summarization and Consolidation 2.1.1 GenAI-Assisted Meeting Minutes Before COVID-19, most corporate meetings were conducted over desk phones. Ever since, they have been hosted over multimedia platforms such as the Zoom, Microsoft Teams, and Google Meet, just to name a few. Unlike the physical desk phones, which cannot be moved arbitrarily, these applications (Apps) allow participation from anywhere using the Internet. Paid corporate accounts are often equipped with even more meeting features. As a result, these Apps have substantially boosted the meeting culture in both industry and academia. Corporate meetings differ significantly from academic ones. In academia, there are mainly two major types: small-scale meetings like a department seminar and large-scale ones such as the annual events of Joint Mathematical Meeting (JMM) or Joint Statistical Meeting (JSM). Novel research discoveries are disseminated through these meetings, and participants are inspired to make potential advancements in their own projects. For academic meetings, there are no hard deadlines or follow-up requirements, e.g., nothing like “ Dr. Jackie Shen - please advance your project “ABCDEFG” by utilizing this methodology “UVWXYZ” as revealed in this meeting by September 4th, 2026 .” Indeed, most scientific advancements could not be possibly prescheduled. This is the key difference from corporate meetings. On corporate floors, meetings are forums where projects are formally kicked off, stakeholders and responsibilities clearly assigned, tasks properly divided and to be conquered, progress tightly tracked and deadlines constantly monitored. Many meetings may also involve stakeholders from all the three Lines of Defense , e.g., Traders, Sales, Analysts, Engineers, Compliance, Legal, Risks, and Internal Auditors. As a result, meeting minutes become critical to track the progress, identify next immediate deliverables, highlight short-term obstacles and solutions, and keep all teams on the same page. They are important for both meeting participants and those who are required but have missed it. In some occasions, they even serve as auditable evidence for regulators and auditors. A meeting mediator or mediating team typically takes quick notes during a meeting, and