Generative AI in Banking and Finance
With the power of GPT 3.5 and large language models, ChatGPT sparks a new change in how the BFSI industry can better enhance its customer experience (CX) and services. From hyper-personalisation to conversational customer support, banks can leverage ChatGPT in a myriad of ways. This article will examine its benefits, risks and potential use cases.
Benefits of ChatGPT
ChatGPT improves CX and the product experience with its NLP (Natural Language Processing) capabilities and ability to automate routine tasks easily. Because of this, agents have more flexibility to focus on other complex tasks which require more care and attention. This not only improves overall efficiency but also enables institutions to direct human expertise to where it is most valuable.
11 Use Cases of ChatGPT
The AI chatbot is a machine when it comes to excelling in both the customer and product experience.
Customer experience
Hyper-Personalisation
- #1 Klarna
Swedish fintech, Klarna, creates a plugin with ChatGPT to offer a highly personalised and intuitive shopping experience. The AI chatbot would then curate recommendations and links to shop in Klarna based on the prompts given by customers.
Source: twimbit analysis
- User: Installs the Klarna extension from the ChatGPT plugin store, then ask for shopping ideas.
- ChatGPT: Provides recommendations based on the conversation
Fraud Detection
- #2 Stripe
Maintaining a robust community on forums such as Discord, Stripe has begun to utilise the advanced capabilities of GPT-4 to detect fraud, ensuring the integrity and security of these communities.
This powerful language model detects fraud by analysing the syntax and context of communications, enabling it to identify potential instances of fraud. In addition, by scanning inbound messages, GPT-4 can effectively flag coordinated activities initiated by malicious actors.
This proactive approach aids Stripe’s ongoing efforts to enhance fraud detection and prevent fraudulent activities.
- #3 Flagright
Flagright has partnered with OpenAI to provide AML compliance and fraud detection capabilities. It functions by training to identify an analyst’s writing style, sentiment and user information. By training this data, Flagright AI can assist in identifying potential fraud.
Conversational Customer Support
- #4 Morgan Stanley
The company is keen on creating an internal-facing chatbot using ChatGPT, exploring how it could enhance insights from financial advisors’ notes and improve follow-up client communications. This is possible due to the company’s vast knowledge base, allowing GPT-4 to provide the best possible advice. As a result, financial advisors intend to use this technology to assist with their client wealth management portfolios.
Exhibit 2: Morgan Stanley’s ChatGPT prototype
Source: twimbit analysis
Product experience
Investment Banking
- #5 JPMorgan Chase & Co.
JPMorgan unveiled a new AI tool that can uncover potential trading signals. This tool uses ChatGPT to analyse the tone of remarks made by policymakers. It aims to aid in identifying potential trading signals to help predict upcoming monetary policy decisions.
Source: twimbit analysis
- #6 Trovata
Trovata AI leverages OpenAI and ChatGPT technology to automate cash workflows and provide business intelligence for corporate finance, accounting & treasury department.
Wealthtech
- #7 Douugh
Douugh is incorporating ChatGPT into its “micro-investing platform”. It enables users to educate themselves about finance. For example, users can look up potential investments, create comparisons, and discover the differences between ETFs and shares.
- #8 Crypto.com
Crypto.com will pilot its first AI-powered user assistant, Amy, supported by ChatGPT to help users learn more about the crypto industry with near real-time information. It will cover specific tokens and projects and details on price listings and historical events.
- #9 OnFinance
OnFinance launched NEO, a ChatGPT-powered chatbot, which helps users learn about finance and investing by answering queries.
SME Banking
- #10 Alaan
Alaan plans to integrate GPT-4 into its user platform. Its purpose is to provide insights into corporate spending and answer critical business questions in real time.
Insurance
- #11 Plum
Plum unveiled PolicyGPT for a more intelligent and user-friendly experience to better educate customers about their health insurance policies.
Source: twimbit analysis
AI-enabled financial services use cases
Brex
Brex launched AI-powered tools for CFOs. The devices use advanced machine learning and NLP technology from OpenAI. It aims to provide relevant insights into corporate spending and answer critical business questions in real time. There would also be a chat interface that will give the users complex, AI-powered insights.
Source: twimbit analysis
Bloomberg
Bloomberg created BloombergGPT, a purpose-built large language model from scratch for finance. This model could potentially provide finance-specific few-shot learning, text generation and conversational systems.
Source: twimbit analysis
ATPBot
ATPBot reduces emotional interference and human error. It analyses real-time market data and uses NLP to extract valuable insights from news articles and other text-based data.
Source: twimbit analysis
Limitations and Challenges
Internal
- #1 Security and Privacy
ChatGPT poses security and privacy risks for banks and customers, requiring diligent monitoring and regulatory compliance.
- #2 Data quality
ChatGPT’s response accuracy relies on data quality and availability, which can be challenging for banks with diverse or outdated data sources.
- #3 Talent Implications
Personnel need training and upskilling to utilise the technology effectively.
- #4 Legal
Organisations must comply with evolving regulations and laws regarding generative AI.
External
- #1 Bias and discrimination
Bias in AI can result from biased training data, flawed algorithms, and unconscious human bias.
- #2 Copyright
Copyright questions about who owns the content once it is on ChatGPT.
- #3 Reputation risk
Insufficient client training may lead to unintentional leaks of sensitive information, negatively impacting the organisation with potential lawsuits.
- #4 Adversarial attacks
Using ChatGPT could expose potential vulnerability to deliberate manipulation by sophisticated external users.
- #5 Cyber Security
Cybercriminals can use ChatGPT to trick them into providing sensitive input with the right prompts.
Future of ChatGPT in Banking
The prospects of ChatGPT in banking are promising. It will revolutionise customer service, enhance operational efficiency, and drive innovation in the industry. With all the benefits that come as ChatGPT continues to evolve, the integration of ChatGPT will reshape financial services, fostering a customer-centric, efficient, and technologically advanced industry.
