Win the Long Game by Prioritizing Data Ethics
Organizations need ethical standards to remain competitive.
Morality isn’t just black and white, especially when it comes to the complex topic of data ethics. This new area of ethical analysis examines how organizations use data by applying the principle of do no harm. In a perfect world, data would be used to benefit consumers, employees and other constituencies.
Data is being used by a range of organizations, from governments to businesses, with each entity having a different set of values. These values should be the starting point in establishing a data ethics program that provides transparency on data use. A well-defined data ethics program enables an organization to decide what does and doesn’t align with its values.
In the last five to ten years, organizations have focused on collecting as much data as possible. Once data is accumulated, it is then used to derive analytics or utilized in a machine learning workflow. The mass accumulation of data can pose a security risk if not managed correctly. And in the case of personal data, it raises privacy and protection concerns.
In my experience, most organizations aren’t prioritizing data ethics, and this is part of the problem. The focus tends to be on how an organization can use data to create value for shareholders or customers. For example, Facebook continuously extracts insights from data to create value for its advertisers, but is it adhering to ethical standards?
Why you need a data ethics program
Organizations that deploy a data ethics program will gain a competitive advantage. These organizations will gain customer trust by demonstrating that they prioritize privacy and can provide full transparency with an ethical data supply chain.
In an ethical data supply chain, an organization tracks data through its entire journey and can explain what’s happening at every stage while ensuring standards are met. It’s the end-to-end journey of data, from collection to destination where, for example, a customer experience is optimized, or data is used to make a business decision.
Organizations that only consider the immediate economic impact of data use are going to fall behind. Enlightened pansophical businesses that have a data ethics program and understand their customers and stakeholders, as well as their obligations to the environment, are more attractive to customers. Similarly, technology workers are looking for organizations that align with their values. Applicants are starting to look for the presence of a data ethics program when they consider working for a company.
Apple has claimed an ethical advantage in relation to its data use by saying that it does not collect personal data. Apple has used this message to mark it as a differentiator from competitors.
Best practices for your data ethics program
Data explainability and an ethical data supply chain are just two elements of a complete data ethics program. Data explainability means providing transparency in the use of data and algorithms. For example, it provides an understanding of how decisions are made by machine learning models, which, in turn, can help to identify bias. The application of data explainability creates the checks and balances to ensure data is being used ethically. It provides visibility into data pipelines to see who is using what data, when they are using it and how they are using it.
And to create an ethical data supply chain, organizations are starting to think about data utilization upfront. What problem is the data trying to solve? Or what customer experience is the data being used to create? Do we need to collect this data set in the first place?
Another consideration is data exposure. Organizations need to establish which pieces of data are being exposed to different users, and how that data Is used. Safeguards must be in place to prevent data access for ethically questionable reasons. For example, an employee of a retail site shouldn’t be able to access and view the buying habits of an individual customer. The need to protect data at the individual level provides a case for the anonymization of data, especially personal data, which can solve many ethical challenges.
How to implement a data ethics program
Organizational buy-in is the first step in establishing a data ethics program. Leadership will need to set the expectation that ethical standards will be maintained in the long term. Employee training must establish acceptable and prohibited actions related to data use. And you’ll need to establish a data ethics board containing representatives from different business functions who can help make decisions on contentious data use cases.
When it comes to individual projects or initiatives, start planning early as it is easier to implement checks and balances at ground zero. This includes mapping out exactly what the data supply chain will look like so that the right data is collected at the right points.
Beware of machine learning bias
Machine learning is becoming increasingly important for organizations to optimize processes or to create additional value, among other things. A key finding from our AI/ML Annual Research Report 2022 was that AI/ML is now the second-most important technology for IT strategy, trailing only cybersecurity. We see examples every day of how prevalent machine learning has become in our lives – from virtual assistants like Alexa and Siri to music recommendations to completing our email texts. Machine learning algorithms are created by people, and even though it is possible to leverage code to augment the process, people are still involved.
With people in the equation, there is potential for bias, whether intended or not. Organizations should take into consideration how machine learning models are constructed to guard against bias. Checks need to be in place. For example, is the data sample large enough? Is the data representative of the population? These types of questions will prevent an organization building inferences into its operations.
When organizations fail to take these factors into account, a lack of an ethical data strategy can have a societal impact. For example, fintech organizations with biased machine learning models might deny a suitable candidate’s credit or loan approval. The bias might be based on factors such as race, gender or socioeconomic status. The correlation may make sense in some cases, but for other cases it could be seen as unfair.
A recent example of this was when Apple partnered with Goldman Sachs to release a credit card. A software developer and his wife provided the same inputs, but due to gender bias, the husband was given a credit limit that was 20 times higher than his wife’s.
The rise of the Chief Ethical Officer
Many big technology organizations, including Facebook and Google, are now employing data ethics teams. And some companies have even hired a Chief Ethical Officer. However, do these teams and executives have the ability to make changes? What is their edict? Or are these roles a reaction to external pressures? At present, it is unclear whether these teams are effective, and it will be interesting to see how these roles develop.
Holding all entities accountable
It is important for customers to ask questions about the use of their personal data. I think people would be surprised to learn all the ways organizations are collecting data from clicks, websites and applications, and potentially even sharing that data with other organizations. Anywhere that you must log in with a password is being tracked. Individuals should be reading data disclosures to understand how organizations leverage data.
By understanding and questioning data use, we can hold all entities accountable. We all need to ensure that organizations are ethical in their use of data as it not only impacts individuals, but society, too.
The Internet of Behavior: What is it and how concerned should we be?
About the Authors
General Manager, Data Services
With over 10 years of experience working with high-growth SaaS/Tech companies, Nihar helps companies achieve their growth objectives through strategic partnerships, corporate development efforts, and executing cross-functional strategic initiatives. He brings significant breadth of experience across strategy, finance, product, and sales and marketing.Read more about Nihar Gupta