About
The rise of Artificial Intelligence (AI) and Machine Learning (ML) has led to calls for companies to manage and use data responsibly and ethically. Scotiabank is taking a leadership position in this area, as one of the first organizations in the financial industry to tackle the challenge of moving beyond declaring principles to operationalizing data ethics – creating tools and processes that address ethical concerns throughout the data and AI lifecycles. The Bank has been ahead of the industry in adopting a public data ethics commitment statement and creating and implementing tools such as the Ethics Assistant (EA) for both Trusted AI and for Trusted Data Use.
Innovation presentation
We live in the age of data. Insights drawn from data, using advanced analytics techniques such as AI and ML are increasingly making up the foundations of decision-making for businesses and governments alike. Simultaneously, we are increasingly becoming aware of the wide-reaching, harmful disparities that have been created because of historical bias and discrimination in information.
Data and analytics have been hailed as objective decision tools that could help us once and for all rid ourselves of human biases and truly treat everyone as equals. These hopes, however, have been shaken by numerous accounts of human bias reflected in data being picked up and amplified, resulting in discriminatory decisions.
These accounts increasingly capture the attention of the public, regulators, and companies, but efforts to mitigate such problems are still in their infancy. The most noticeable evidence of this is the growing focus on AI ethics, and the risk surrounding the development and use of AI/ML models including algorithmic bias. Even so, according to a recent IBM survey (https://www.ibm.com/downloads/cas/4DPJK92W) a quarter of responding organizations have operationalized AI Ethics. This is while a Deloitte Canada survey found that 90% of global consumers would sever ties with an organization if the company used their data unethically.
The ethical risks we face as we are increasing the use of data and analytics are not just related to AI/ML-- there are significant ethical considerations that exist throughout the data lifecycle, from the moment data is collected to when it is eventually deleted.
For that reason, Scotiabank has chosen to focus beyond just the ethics of AI/ML and is one of the first organizations in the financial industry to tackle the ethical concerns that can arise throughout the entire data lifecycle. It is also one of the first financial institutions to operationalize data ethics as a centralized function and at scale.
In centralizing the function, Scotiabank’s Data Ethics team acts as an internal centre of excellence for data ethics. Leveraging this position within the organization, the team has created tools for ensuring appropriate data ethics considerations are brought to practitioners’ attention at the right time in their work, and that a robust governance process exists to ensure data ethics decisions and trade-offs are made with the outmost care. Additionally, the team has been a driving force in fostering and building a culture of ethical data use through ongoing education efforts. As a result, there is Bank wide understanding that all employees are accountable for the ethical use of data and there is a common understanding of what ethical data use means. Teams across the Bank are aware of the supports in place to ensure ethical considerations are prioritized.
As an early adopter of Data Ethics, Scotiabank has incorporated leading ethical practices, including developing one of the first public Data Ethics commitment statements (https://www.scotiabank.com/ca/en/about/responsibility-impact/with-our-c…) in the financial industry, incorporating data ethics into the Bank’s Code of Conduct, and creating and implementing tools such as the Ethics Assistant (https://www.scotiabank.com/ca/en/about/perspectives.articles.digital.20…) for both Trusted AI and for Trusted Data Use. In addition to these activities, the Bank is developing a mandatory training course to ensure data and analytics practitioners across the Bank understand Scotiabank’s Data Ethics principles and tools.
Scotiabank’s Ethics Assistant, (https://www.newswire.ca/news-releases/scotiabank-launches-ethics-assist…) draws attention to the ethical considerations that must be acknowledged during the development and design process for analytics models. Currently all new AI/ML models at the Bank that use customer data across Canada must complete the Ethics Assistant (EA) tool, with the intention to expand this scope globally.
Similarly, the Trusted Data Use (TDU) tool evaluates what ethical considerations should be made when using individual customer data to make business decisions that directly impact the customer. These tools, combined with resources and education campaigns created by the Bank’s designated Data Ethics team, position Scotiabank’s AI & data ethics program as one of the most comprehensive programs in the financial industry.
Uniqueness of the project
Scotiabank’s focus on ethics at every stage of the data life cycle is unique across Financial Services. While many organizations are raising awareness around the ethical considerations behind AI models, Scotiabank’s broad approach to data ethics is a differentiator. Institutions who have taken measures in this space often focus primarily on AI ethics, without a dedicated ethics team. Below outlines some of the ways Scotiabank has taken a leadership position in this area.
- Dedicated Data Ethics team: In 2020 Scotiabank created a designated Data Ethics team to foster a culture of the ethical use of data at the Bank. Today the team includes 9 individuals focused on design and delivery, governance and operations, and education and partnerships. The team works to develop and implement effective tools to assist the business in developing ethical algorithms and treating data ethically. To our knowledge, Scotiabank was the first Canadian Bank with a designated Data Ethics team.
- Public Data Ethics Commitment: Scotiabank’s public Data Ethics Commitment guides the Bank in its efforts to ensure that it is adhering to its principles around the ethical use of data. The principles outline a commitment to the use of and access to data in a useful, adaptable, accountable, transparent, respectable, and safe manner.
- Data Ethics as part of the Bank’s Code of Conduct: To ensure that all employees remain accountable to act in accordance with the Bank’s data ethics principles, the principles have been incorporated into the Bank’s Code of Conduct, which employees attest to on an annual basis.
- Data Ethics training: To build a culture of ethical data use at the Bank, the Bank’s Data Ethics Office offers live training sessions, and is developing a mandatory training course for data and analytics practitioners, that will be launched in the new year. Training is supplemented with written resources highlighting the importance of action in this space and outlining how data ethics can be implemented at different stages of the data lifecycle.
- Partnerships: To ensure the highest quality of the Bank’s program and alignment to leading industry standards, Scotiabank has partnered with several organizations to help execute its data ethics strategy. Most prominently Scotiabank partnered with IEEE and Queen’s University on the delivery of a Data Ethics certification, the IEEE Trusted Data and AIS playbook.
- Presentations/Articles/Forums: Scotiabank is an active participant in industry forums such as the Data for Good Global Summit (https://www.cdomagazine.tech/cdo_magazine/news_feed/videos/dataforgood_…) and been featured in podcasts (https://open.spotify.com/episode/0oIqDHHeQlBgxQjTaedwBS) and articles (https://bankautomationnews.com/allposts/retail/inside-look-scotiabanks-…) raising awareness around the practice.
- Tools: Scotiabank has developed tools and processes (https://www.scotiabank.com/corporate/en/home/media-centre/media-centre/…) for the incorporation of Data Ethics considerations throughout the data lifecycle. These tools enable a large-scale, consistent, systematic review of data ethics-related risks across the Bank and bring proactive guidance to practitioners at key points in their work with data and analytics.
