The Trustworthy AI Podcast

In this episode, we talk to Anant Bhardwaj, who founded and leads Instabase, an AI application platform for unstructured data. Instabase has been helping enterprise customers worldwide streamline their operational processes, by using AI and workflow automation to intelligently process documents. 

Anant talks about the evolution of the underlying language/document processing techniques underpinning Instabase—from program synthesis to deep learning, and, more recently, large language models (LLMs). He shares his nuanced view on LLMs: bullish on its broader promise, but cautious about its near term usability in enterprise use cases due to issues around lack of predictability and security/privacy concerns. 

In this episode, we talk to Tobias Baer, author of Understand, Manage and Prevent Algorithmic Bias, perhaps one of the earliest books (2019) on Trustworthy AI. 

We cover a range of topics - from bias in algorithms and practical ways organizations can address it;  what data scientists in a financial institution can learn from Greek literature, German philosophers, and martial arts; and finally, how AI is viewed differently in Asia and the West.

About the speaker

Tobias Baer, based in Taiwan, is a pioneer in data science and the application of analytics and psychology to risk management in banking. After more than two decades at McKinsey, he now supports a select set of Fintechs/start-ups as Senior Advisor and pursues psychological research in cooperation with the University of Cambridge.

In episode #2, we first explore the world of Money Laundering, Terrorist Financing and other types of financial crime, and understand how financial institutions use data and analytics  to prevent and detect such crime. 

We discuss the importance of Trustworthy AI in this regulated part of the industry and finish with a story about data science and poetry. Tune in to discover why building models to mitigate financial crime may just be one of the coolest jobs in data science. 

About the speaker

John Cusack is Chair of the Global Coalition to Fight Financial Crime. If you have worked in the financial crime compliance space in banking, chances are that you have probably worked for John, or know someone who has. Among other things, John has played the role of Global Head of Financial Crime Compliance at two global banks (UBS and Standard Chartered) and has been a two-time Co-Chair of the Wolfsberg Industry Group for the management of financial crime risks.

In episode #3, we are joined by two guests, Brad Carr and Natalia Bailey from the Institute of International Finance (IIF) to learn about some of the most comprehensive research conducted in the adoption and governance of AI in the Finance industry, over the last 3 years. 

With more than 450 members from 70 plus countries, the IIF's mission is to support the industry in the management of risks and to advocate for appropriate regulatory, financial, and economic policies. The IIF’s annual Machine Learning surveys since 2018 has provided invaluable insight into AI adoption, barriers and governance in the industry.

We also discuss the IIF’s Data Ethics Charter, and explore why banks have an opportunity, and perhaps a duty, to make trust a key part of their customer proposition.

About the speakers

Brad Carr is Managing Director for Digital Finance at the IIF and has led the IIF’s work across a range of data and digital topics. Before joining the IIF in 2014, Brad spent over a decade in National Australia Bank. 

Natalia Bailey is a policy advisor in the same team and has played multiple roles at the IIF over more than a decade. Natalia is also the driving force behind a lot of the IIF’s work around AI governance and data ethics.

In Episode #4, we discuss the importance of high quality external data in enabling greater AI innovation and adoption. We explore how data management best practices around consent, vendor due diligence and data quality monitoring can help build trust in such data. And how, when used appropriately, more external data can help make AI models fairer and more reliable. Finally, we get Mark’s take on why the financial services industry remains one of the best places for anyone interested in data and AI. 

About the speaker

Mark Hookey is the CEO of Demyst, an external DataOps platform that provides accelerated deployment of external data solutions for the world’s leading banks, insurers and fintechs. Demyst’s platform eliminates the friction associated with data procurement, testing, and deployment to unlock innovation and drive business growth.

In episode #5, we discuss the role that good regulation can play in promoting the adoption of AI at scale. David shares his experiences in the transition from writing the finance industry’s first-ever regulatory guidelines on AI (at the Monetary Authority of Singapore, 2018) to actually putting them into practice.

We get David’s views on whether AI is living up to its promise in financial services, and the skill sets needed to do the same, and finish by peering into the future and getting his predictions on where AI can have the maximum impact in the next 3-5 years.

About the speaker

David Hardoon is Managing Director for Data Innovation at the Aboitiz Group (one of the biggest conglomerates in the Philippines) and heads Data and AI at the Group’s banking arm, the Union Bank of Philippines. Prior to this, David was the Chief Data Officer at the Monetary Authority of Singapore, where he authored the world’s first-ever regulatory guidelines on AI in financial services in 2018. He holds a Ph.D. in Machine Learning and is widely recognized as one of the pioneers of “trustworthy AI” in financial services.

In episode #6, we talk to Ansgar Koene, Global AI Ethics and Regulatory leader at EY. Ansgar, based in London, focuses his work on tools to maximize the beneficial use of technology and minimize negative consequences on people and society. We get Ansgar’s take on AI Ethics, and how that varies, or doesn’t, from Ethics in business more broadly.

We get an expert’s perspective on the draft EU AI law, proposed in April last year; a consultant’s perspective on how his clients worldwide are building up their Trustworthy AI capabilities, and an academic’s view on making the transition to industry.

About the speaker

Ansgar Koene has a multi-disciplinary research background, having worked and published on topics such as Algorithmic Policy and Governance, data privacy, AI Ethics and Standards, bio-inspired Robotics, and experimental Human Behaviour/Perception studies. Ansgar holds an MSc in Electrical Engineering and a Ph.D. in Computational Neuroscience.

In episode #7, we talk to Peter Kasahara, Group Chief Data Officer at Zurich Insurance Group, about the AI opportunity in insurance. We discuss the barriers to AI adoption in the industry, and the importance of trust in that context. We discuss the attractiveness of the insurance industry for data professionals. Finally, Peter shares his views on the AI opportunities about which he is most excited.

About the speaker

Peter leads global data management and governance at Zurich Insurance Group, and is responsible for developing new data analytics capabilities and products, expanding Zurich’s machine learning competencies, and ensuring delivery against Zurich’s data for good commitments. Before joining Zurich, Peter was Managing Partner at PwC’s Digital Intelligence practice, and also built the data and analytics division of IBM in Europe. Peter is based in Zurich, Switzerland.

In episode #8, Oliver Maspfuhl provides us a European perspective on a range of topics - from the new draft AI regulation proposed last year, to the state of AI adoption in Europe, and how organizations are building practical mechanisms to make AI trustworthy.

Oliver also shares his advice for data scientists who are weighing a career in financial services against other options, and provides his predictions on the impact of AI in the next few years.

About the speaker

Oliver Maspfuhl is Chapter Lead Data Scientist in the central Big Data & Advanced Analytics unit of Commerzbank AG, focusing on developing Trustworthy AI applications for making banking a data-driven business, mostly in Risk Management and Financial Crime Prevention.

In episode #10, we are joined by two guests, Andy McMahon and Greig Cowan from the NatWest Group, to learn about the group’s Machine Learning Operations (MLOps) journey.

Andy and Greig talk about the four pillars of that journey - People, Process, Data and Technology - and about the efforts to embed ML model quality by design into those pillars. They explain how the group’s standardised data lab methodology, supported by a modern toolset, has enabled the group to scale its data science capability rapidly.

About the speakers

Andy McMahon heads MLOps at NatWest Group, while Greig Cowan heads Data Science and Engineering for Data Innovation at the group. Both have extensive experience of taking analytics solutions from ideation to production. 

Andy’s experience spans model building, software development and technical management. His particular area of interest is the challenge of deploying machine learning based software products at scale. He is also the author of a book on ML engineering.

Greig has over 15 years of experience at the leading edge of scientific research, including with the Large Hadron Collider, where he developed physical models to statistically extract fundamental parameters from large datasets. He now focuses on data science innovation within the financial services industry.