In this interview, we sit down with Swati Tyagi, Senior Applied AI/ML Researcher, to delve into her journey through the world of artificial intelligence and machine learning. With a unique background that bridges academia, finance, and healthcare, Swati shares her perspective on how AI is reshaping industries, from revolutionizing financial processes with large language models to addressing the complexities of ethical AI. Her insights offer a glimpse into the evolving role of AI in driving innovation while navigating the regulatory landscape. Read on for valuable lessons on AI’s transformative potential in high-stakes industries.
Explore more interviews like this here: Tommy Tran, Software Engineer at Ramp — Transitioning from Game Development to AI, Measuring Impact, Balancing Innovation, Introducing AI in SMEs, and AI’s Role in U.S. Business and Economy
What inspired your specialization in AI/ML, and how did your academic background shape your career path?
My passion for artificial intelligence and machine learning was ignited during my undergraduate studies in computer science, where I became fascinated by the capacity of intelligent systems to tackle and solve complex real-world challenges. Earning a bachelor’s degree in computer science and engineering further fueled this interest, motivating me to pursue a PhD. This advanced education allowed me to deeply explore Artificial Intelligence/Machine Learning(AI/ML), GenAI, large language models (LLMs), and the principles of ethical, secure, and responsible AI. Throughout my academic journey, I developed a strong foundation in key areas such as machine learning, algorithms, data structures, convex optimization, and statistical analysis, all of which are essential for creating advanced and reliable AI/ML models.
Additionally, engaging in interdisciplinary collaborations across healthcare, technology, and finance exposed me to a diverse array of AI applications, inspiring me to bridge the gap between theoretical research and practical industry solutions. This combination of rigorous academic training and a commitment to impactful technology has been instrumental in shaping my career, leading me to a specialized role in Artificial Intelligence/Machine Learning within prominent financial institutions like JPMorgan Chase.
How are large language models revolutionizing traditional financial processes, and what are their most significant applications at JP Morgan Chase?
Large language models (LLMs) are fundamentally transforming traditional financial processes by enhancing data analysis capabilities, improving customer interactions, and automating intricate tasks. At JPMorgan Chase, LLMs play a pivotal role in several key areas. In customer service, natural language processing (NLP) powered by LLMs enables chatbots and virtual assistants to deliver instant, accurate responses to inquiries, thereby elevating the customer experience and reducing response times. In risk management and compliance, LLMs analyze vast amounts of unstructured data, including regulatory documents and transaction records, to identify potential risks and ensure adherence to financial regulations. Additionally, LLMs contribute significantly to fraud detection by processing complex transaction narratives to identify suspicious activities and patterns indicative of fraudulent behavior.
In financial forecasting and insights, LLMs assist in generating detailed financial reports and predictive analyses by interpreting market trends and economic indicators, facilitating more informed decision-making. The integration of an LLM Suite is also shaping the operational strategies for each employer within the organization. Collectively, these applications not only streamline operations but also enable more proactive and strategic approaches to managing financial services.
What unique challenges and opportunities arise when integrating AI, statistics, and finance in real-world projects?
Integrating AI, statistics, and finance in real-world projects presents a unique blend of challenges and opportunities. One of the primary challenges is ensuring data quality and availability, as financial data is often sensitive, proprietary, and subject to stringent privacy regulations, making access and preprocessing difficult. Model interpretability is another significant hurdle, as financial decisions demand a high level of transparency and explainability, which can be challenging to achieve with complex AI models. Additionally, regulatory compliance is critical, requiring that AI-driven solutions adhere to stringent financial regulations and standards, necessitating continuous monitoring and adaptation. Integrating advanced AI models with legacy financial systems can also be technically demanding and resource-intensive. Moreover, while large language models (LLMs) offer promising advancements in processing and analyzing vast amounts of unstructured data, they come with their own set of challenges. Issues such as model hallucinations, where AI generates inaccurate or misleading information, and the need for secure systems to protect sensitive financial data are significant concerns that must be addressed to ensure reliable and safe AI implementations.
On the opportunity side, the integration of AI and statistical models enhances decision-making by providing deeper insights and more accurate predictions, enabling better strategic choices. Operational efficiency is significantly improved through the automation of routine tasks, which reduces costs and minimizes human error, allowing professionals to focus on more value-added activities. AI also facilitates personalized financial services, enabling the customization of financial products to meet individual customer needs, thereby improving satisfaction and loyalty. Furthermore, the integration of AI opens the door to innovative financial products and services that were previously unimaginable, fostering competitive advantage and market growth. By effectively addressing these challenges and leveraging the opportunities, organizations can substantially enhance their financial operations and gain a competitive edge.
How do you approach the development of ethical AI frameworks, particularly in addressing bias within financial systems?
Developing ethical and responsible AI frameworks, especially to address bias within financial systems, requires a comprehensive and multi-faceted approach. The first step involves rigorous data auditing and preprocessing to ensure that the training data is representative and free from historical biases. This process includes thorough auditing, cleaning, and preprocessing to eliminate any biased patterns that could influence the model. Implementing fairness metrics, fair lending practices and continuous monitoring is crucial to assess and mitigate bias in AI models, utilizing techniques such as disparate impact analysis and equal opportunity metrics to ensure equitable outcomes.
Transparent model design is another critical component, involving the creation of interpretable and explainable models that allow stakeholders to understand the decision-making process. This transparency is essential for building trust and accountability. Additionally, fostering inclusive development teams by promoting diversity ensures that varied perspectives are considered, reducing the likelihood of unconscious biases influencing model design. Adhering to regulatory compliance and industry best practices for ethical AI is also vital, requiring ongoing updates and adherence to evolving standards to maintain integrity.
Continuous monitoring and feedback loops are established to promptly address any emerging biases or ethical concerns in AI systems once they are in production. By embedding these principles into the AI development lifecycle, we can create financial systems that are not only effective but also fair and trustworthy, ensuring equitable access and maintaining stakeholder confidence.
What role do you see AI and automation playing in the evolution of business operations over the next decade?
Over the next decade, artificial intelligence (AI) and automation are poised to revolutionize business operations across multiple dimensions, driving unprecedented levels of efficiency and innovation. Leveraging advanced AI agents and generative AI (GenAI), businesses will achieve significant operational enhancements by automating repetitive and time-consuming tasks. This will streamline workflows, reduce costs, and substantially increase productivity, allowing organizations to allocate resources more strategically and focus on high-value, innovative activities will transform data-driven decision-making through the deployment of large language models (LLMs) and GenAI, enabling the rapid and precise analysis of vast datasets. These technologies will generate actionable insights that empower informed and timely strategic decisions, enhancing overall business agility.
In customer experience, AI-powered personalized interactions, facilitated by LLMs and GenAI, will lead to heightened customer satisfaction and loyalty by enabling businesses to anticipate and respond to individual customer needs with exceptional accuracy. Innovation and the development of new business models will be catalyzed by AI, allowing for the creation of cutting-edge products and services that were previously unattainable. This will foster competitive advantage and drive significant market growth. Additionally, AI will play a crucial role in risk management and cybersecurity by enhancing risk assessment, fraud detection, and the implementation of sophisticated AI-driven security measures to safeguard businesses against evolving threats. Furthermore, AI and automation will drive a fundamental transformation of the workforce by augmenting human capabilities. Employees will be empowered to focus on tasks that require creativity, critical thinking, and emotional intelligence, while AI handles routine and data-intensive operations. This synergy between human talent and AI technologies will cultivate a more dynamic and innovative organizational culture.
Overall, AI and automation, including GenAI and LLMs, will be integral to building agile, resilient, and forward-thinking organizations capable of thriving in an increasingly complex and dynamic business landscape. These technologies will not only optimize existing processes but also unlock new avenues for growth and excellence, fundamentally reshaping the future of business operations.
Can you share insights into the techniques you’ve developed to enhance fairness and transparency in AI-driven credit scoring models?
Enhancing fairness and transparency in AI-driven credit scoring models is paramount to ensuring equitable access to financial services. To achieve this, I have developed and implemented several key techniques. First, bias mitigation algorithms such as reweighting, resampling, and adversarial debiasing are incorporated during the model training process to minimize biases related to sensitive attributes like race, gender, and socioeconomic status. Explainable AI (XAI) techniques, including SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations), are utilized to provide clear and understandable explanations for credit decisions, ensuring that stakeholders can comprehend how scores are determined.
Additionally, fairness constraints are embedded directly into the optimization process of machine learning models to ensure that predictions do not disproportionately disadvantage any particular group. Comprehensive feature selection is carefully conducted to avoid proxies for sensitive attributes, thereby reducing the risk of indirect bias in the model’s predictions. Regular auditing and validation of credit scoring models are performed to detect and address any emerging biases or disparities in performance across different demographic groups. Engaging with diverse stakeholders, including ethicists, legal experts, and community representatives, ensures that a wide range of perspectives are incorporated into the model development process, thoroughly addressing ethical considerations. Transparent reporting practices are maintained, providing detailed documentation that outlines the methodologies, data sources, and fairness assessments used in developing credit scoring models. This transparency builds trust with both internal and external stakeholders. By implementing these techniques, I strive to create credit scoring models that are not only accurate and efficient but also fair and transparent, fostering trust and ensuring equitable access to financial opportunities for all individuals.
How does your experience in healthcare inform your approach to AI applications in the financial sector?
My experience in the healthcare sector, particularly during my tenure at Evernorth, has profoundly shaped my approach to applying AI in the financial industry by emphasizing the critical importance of data integrity, ethical considerations, and transparency. At Evernorth, I invented an AI graph-based algorithm designed to address customer issues promptly by accurately detecting their key problems within system interfaces. This innovation not only enhanced customer satisfaction by ensuring timely and effective problem resolution but also underscored the necessity of robust data governance and compliance with stringent regulatory standards. In healthcare, AI systems must adhere to rigorous regulations and protect patient data privacy, instilling in me a meticulous approach to data governance and compliance. These principles seamlessly translate to the financial sector, where safeguarding sensitive financial information and adhering to regulatory requirements are equally paramount. Additionally, the interdisciplinary collaboration inherent in healthcare AI projects at Evernorth equipped me with the ability to work effectively with diverse teams, including domain experts, policymakers, and technologists. This collaborative mindset is essential in finance, where integrating AI requires a deep understanding of complex financial instruments and regulatory landscapes.
Moreover, addressing biases and ensuring fairness in healthcare AI has heightened my awareness of similar challenges in financial applications, such as credit scoring and fraud detection. This experience allows me to implement robust bias mitigation strategies that promote equitable outcomes, ensuring that AI-driven financial solutions are both effective and fair. By leveraging the insights and skills gained from healthcare AI projects, I am able to develop and deploy AI technologies in the financial sector that prioritize ethical standards, transparency, and the protection of sensitive data, ultimately fostering trust and driving innovation.
What are your key strategies for fostering innovation while navigating the regulatory complexities of the finance industry?
Fostering innovation within the regulatory framework of the finance industry necessitates a balanced approach that harmonizes cutting-edge technological advancements with strict compliance standards. My key strategies to achieve this include proactive compliance integration, continuous learning and adaptation, the implementation of ethical AI frameworks, collaborative partnerships, and robust risk management. Proactive compliance integration involves embedding regulatory requirements into the AI development lifecycle from the outset, ensuring that innovative solutions are designed with compliance in mind, reducing the risk of regulatory setbacks, and facilitating smoother implementation. By integrating compliance early, we can anticipate and address potential regulatory challenges before they impede progress. Continuous learning and adaptation are crucial for staying abreast of evolving regulations and industry standards, which involves regular training, attending industry conferences, and collaborating with regulatory bodies to understand upcoming trends and requirements.
Implementing ethical AI frameworks that prioritize fairness, transparency, and accountability ensures that AI applications adhere to ethical standards that often align with regulatory expectations, fostering responsible innovation. Collaborative partnerships with legal experts, compliance officers, and industry stakeholders create an environment where innovative ideas can be vetted for compliance early in the development process, enabling agile responses to regulatory challenges and ensuring that innovations are both compliant and impactful. Finally, robust risk management involves developing and maintaining comprehensive risk management protocols, including thorough testing and validation of AI models, to identify and mitigate potential compliance risks before they escalate. By integrating these strategies, I drive innovation that not only pushes the boundaries of what AI can achieve in finance but also respects and adheres to necessary regulatory frameworks, ensuring sustainable and responsible growth.
How do you envision the future of ethical AI shaping global finance, and what steps are necessary to achieve widespread adoption?
The future of ethical AI in global finance is poised to transform the industry by enhancing transparency, fostering trust, and promoting inclusive financial practices. Ethical and Responsible AI will drive innovations that optimize financial operations while ensuring that these advancements benefit a broader spectrum of society. Key aspects of this transformation include enhanced transparency and trust, inclusive financial services, and sustainable financial practices. Enhanced transparency and trust will be achieved as ethical AI prioritizes explainability, allowing stakeholders to understand and trust AI-driven decisions. This transparency is crucial for building confidence among customers, regulators, and financial institutions, ensuring that AI systems are accountable and their decision-making processes are clear. Inclusive financial services will be promoted by mitigating biases and ensuring fairness, providing equitable access to credit, investment opportunities, and personalized financial advice to underserved populations. Ethical and Responsible AI will help bridge gaps in financial inclusion, ensuring that all segments of society can benefit from financial innovations. Sustainable financial practices will be supported by ethical AI through better risk assessment, promoting responsible investing, and ensuring that financial products align with environmental, social, and governance (ESG) criteria.
This alignment will drive the financial sector towards more sustainable and socially responsible practices, contributing to broader societal goals. To achieve widespread adoption of ethical AI in global finance, establishing global standards, securing regulatory support and incentives, raising education and awareness, fostering collaborative ecosystems, and implementing robust governance frameworks are essential. Developing and harmonizing international ethical AI standards ensures consistency and facilitates cross-border collaboration. At the same time, clear guidelines and incentives from governments and regulatory bodies encourage financial institutions to prioritize ethics alongside innovation. Raising awareness among stakeholders and fostering collaboration between financial institutions, technology providers, academia, and regulatory bodies create a supportive ecosystem for ethical AI solutions. Additionally, strong governance frameworks that oversee AI development and deployment ensure ongoing adherence to ethical standards and facilitate continuous improvement. By taking these steps, the financial industry can harness the full potential of AI while ensuring that its applications are ethical, transparent, and beneficial to society at large, ultimately building a more equitable and trustworthy global financial system.
What guidance would you offer to emerging AI researchers aiming to drive impactful change in high-stakes industries like finance and healthcare?
For emerging AI researchers aspiring to drive impactful change in high-stakes industries such as finance and healthcare, prioritizing ethical considerations is paramount. Researchers should always place ethics at the forefront of their work, understanding the societal implications and striving to develop AI solutions that promote fairness, transparency, and accountability. Developing deep domain expertise in the specific industry is crucial, whether it’s familiarizing oneself with financial instruments and regulatory requirements in finance or understanding clinical practices and patient privacy laws in healthcare. Fostering interdisciplinary collaboration by engaging with professionals from diverse backgrounds, including domain experts, ethicists, policymakers, and technologists, leads to more holistic and robust AI solutions. Embracing continuous learning ensures that researchers stay updated with the latest advancements, methodologies, and regulatory changes, maintaining the relevance and effectiveness of their work.
Focusing on explainability and transparency in AI models builds trust among users and stakeholders, which is essential in high-stakes environments. Committing to responsible data practices by handling data responsibly, respecting privacy and security standards, and mitigating biases promotes equitable outcomes and trustworthy AI systems. Engaging with regulatory bodies helps researchers design compliant and sustainable AI solutions while aiming for real-world impact ensures that their work addresses tangible societal needs. Seeking mentorship and networking opportunities accelerates growth and opens avenues for collaboration, and committing to lifelong ethics education ensures that researchers remain aligned with evolving societal values and standards. By following this guidance, emerging AI researchers can effectively contribute to transformative and ethical advancements in finance, healthcare, and beyond, ensuring that their work not only drives innovation but also upholds the highest standards of responsibility and integrity.
