Presentation on Big Data & Algorithms in (Fair) Consumer Finance at Computers, Privacy and Data Protection (CPDP 2017) Panel on ‘Privacy, Data Protection, and the Governance of Algorithms’ https://youtu.be/akEg3Qrkbl0
Catherine G. van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity Advisor, Researcher & Lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the Faculty of Public Management, Law and Security - The Hague University of Applied Sciences.
Former expert member of the Financial Services User Group(FSUG)-European Commission T: @CatherineGvh
How AI, OpenAI, and ChatGPT impact business and software.
Big data algorithms in fair consumer finance CPDP2017
1. Big Data & Algorithms in (Fair)
Consumer Finance
Catherine Garcia van Hoogstraten, Digital Governance, Information
Technology Law & Policy & Cybersecurity Advisor, Researcher &
Lecturer at the Cybersecurity Program -Webster University(Leiden)
Former expert member of the Financial Services User Group(FSUG)-
European Commission
T: @CatherineGvh
George Herbert Walker
School of Business & Technology
CPDP 2017 Panel on ‘Privacy, Data
Protection, and the Governance of
Algorithms’
https://youtu.be/akEg3Qrkbl0
2. Big Data and Algorithms in Consumer Finance
Potential Impacts & Benefits
Speed up access to financial P’s & S’s
Financial advice guidance
Budgetary management
Preventing over-indebtedness
Identify fraud, money-laundering, bribery and other
corrupt practices
FSUG-EC (June 2016) Assessment of current and future
impact of Big Data on Financial Services
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
3. Detrimental effects of automated decisions applied
in algorithmic creditworthiness assessment requires
adaptive regulation:
Accurate predictability of future behavior
Risk profiling assessment
Algorithmic decision making transparency
Biased or discriminatory outcomes
3
Big Data and Algorithms in Consumer Finance
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
4. Automated Decisions applied in Algorithmic
Creditworthiness Assessment
Conditions that must be met for the Art. 22(1) GDPR to
apply:
1. there must be a decision;
2. that decision is based solely on automated data
processing;
3. the data processed for the decision are intended to
evaluate certain personal aspects (creditworthiness) of
the person concerned;
4. the decision must have legal or other significant
effects for the person(unfair discrimination due to the
denial of credit)
4
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
5. Sources of Bias in the Algorithmic System
Input
Training
Programming
Algorithms that use Big Data techniques for
underwriting consumer credit can be vulnerable
to all three of these types of bias risks.
5
White & Case (2017) Algorithms and bias: What lenders need to know
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
6. Impact of Bias in the Algorithmic Creditworthiness
Assessment
Big Data analytics based in nontraditional data
Nontraditional data sources: databases containing
internet search histories, shopping patterns, social
media activity etc
Denial is the output from an opaque algorithm
analyzing nontraditional data= unfair or
discriminatory lending decisions
Challenges in translating algorithmic decisions when
large potential borrowers are denied credit.
6
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
7. The right to learn the logic behind processing
Right of access by the data subject Art. 15(h)
GDPR about:
The existence of automated decision-making
Meaningful information about the logic
involved and
The significance and consequences of such
processing for the data subject.
7
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences . Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
8. Legal and Tech challenges ahead
Profiling and monitoring of individuals by automated
machines
Transparencyof automated decisions
Supervision of automated decisions taken by
autonomous machines when collecting personal data
Accountability of the algorithm design
Test and monitor the algorithms
Maintaining human oversight, comprehension and control of
emerging AI systems should be at the core of the technology
8
Catherine Garcia van Hoogstraten, Digital Governance, Information Technology Law & Policy & Cybersecurity
advisor, researcher & lecturer at the Cybersecurity Program -Webster University(Leiden). Lecturer at the
Faculty of Public Management, Law and Security - The Hague University of Applied Sciences. Former
expert member of the Financial Services User Group(FSUG)- European Commission T: @CatherineGvh
George Herbert Walker
School of Business & Technology