Panel Discussion: The AI Effect on Credit Worthiness

On September 9, 2021, senior practitioners and decision makers in the financial services industry came together at the ForHumanity AT->GO Conference to examine the shortcomings and opportunities of automating the credit provisioning process. Quite a lot was covered in ninety minutes, so for your on-demand convenience, we have broken it down into twenty bite sized pieces. Enjoy!


	PANEL PARTICIPANTS: Paul Lashmet (Moderator) - ForHumanity Fellow and Managing Director at North Castle Integration Marino Vedanayagam - Head of Risk, Business Governance and Controls, Institutional, Americas at ANZ Bank Nick Curcuru - VP Advisory Services at Privitar Peter Memon - Head of Analytics at FactSet Ted Frisch - Strategic Sales at Fiddler AI

View as YouTube Playlist

Featured

Panel Discussion Preview: The AI Effect on Credit Worthiness

Automated credit decisioning is designed for scale, efficiency and, in a world of blue skies and sunshine, to reduce subjective bias from the process. But the world is not perfect.

Determining the credit worthiness of an individual, a small business, or a corporate entity has real-world impacts that cannot be taken lightly. While accountability, governance and oversight is necessary, the safeguards we employ should not impair progress.

Episode 1: Opening Discussion About Bias: Good and Bad

Often, conversations on the topic of bias and AI are primarily concerned with prejudicial or intolerant decision making processes hidden within the algorithms. That is a real concern, but it limits discourse to a binary debate of AI = Good OR Bad.

In this conversation about bias, our panelists take a nuanced and neutral approach. The algorithms that power AI are not inherently biased because math, on which algorithms are built, is not inherently biased. Bias, good and bad, is introduced through data, strategy, explainability, and projected expectations.

Episode 2: Transparency Begets Accountability and Trust

Accountability and trust is what ForHumanity is all about. This discussion posits that it all starts with transparency. If we are able to pull back the curtain on AI models to better understand how machine learning algorithms are built, then we will go a long way towards alleviating the apprehensiveness that people rightfully have with regard to what is hidden inside the black box.

It is incumbent on us to explain what is going on vs. defaulting to a machine that takes over. The black box is no longer an excuse because our regulators, our peers, and our customers are asking a lot of good questions, such as: What was the intent of the AI model? What was put inside the model? Are the outcomes just? How is that measured? - and more.

Episode 3: The Need for Human Review

Banks build engines to help automate the credit provisioning process, suggesting what positions to take as it relates to lines of credit or other credit products that are being offered. But these engines are built by engineering and credit modelling teams, not the bankers, relationship managers, and other customer and regulatory facing teams. Since they don’t know the details of inputs, logic, and outputs, they are often at a loss as to why a certain position was taken.

It is incumbent on them, the people who own the products, to understand the basics of engineering and AI in order to understand how these positions are driven, not just to explain but to review, challenge, and verify. There always needs to be a human element.

Episode 4: Technology to Help Explain What's Going On

In Episode 3, we conveyed the need for human review in all things AI with some examples. In this episode, we discuss how new applications of technology enable us to do that effectively through transparency. With that, we can explain and show how data is sourced, created, transformed, consumed, and presented (by what and by whom) as it moves through our systems.

Today, there is a lack of transparency. Organizations do not adequately monitor and understand the algorithms and functions that are making critical business decisions on the company’s behalf and their client’s behalf.

As we advance and rely more on artificial intelligence and autonomous systems, we need to be able to build the storyboard so that, at any point of time, we (humans) can explain what is going on.

Coming Soon! Episode 5: Are We Doing It Right?

ForHumanity is a tax exempt public charity that is establishing audit rules, standards, guidelines and where appropriate, public policy for all Artificial Intelligence (AI) and autonomous systems in the areas of Ethics, Bias, Privacy, Trust, and Cybersecurity.

Licensing Information
ForHumanity grants North Castle Integration permission to display AT->GO Conference recordings on its website and links back to the https://ForHumanity.center website. This limited license is granted for such time as either party may choose to either revoke permission or discontinue displaying the recordings.