Artificial intelligence has the power to transform decision-making processes, but it can also deepen existing biases if not carefully managed. In the sections below, we invite you to explore examples of long-term decisions that can be influenced by AI, showing how biases embedded in data may affect outcomes with lasting consequences. As discussed in "Unequal Roots, Unequal Outcomes: The Deepening Bias in Modern AI and How to Uproot It," understanding and addressing these biases is crucial to ensuring that AI supports fair and equitable decision-making.
About the author: Jeff Hulett leads Personal Finance Reimagined, a decision-making and financial education platform. He teaches personal finance at James Madison University and provides personal finance seminars. Check out his book -- Making Choices, Making Money: Your Guide to Making Confident Financial Decisions.
Jeff is a career banker, data scientist, behavioral economist, and choice architect. Jeff has held banking and consulting leadership roles at Wells Fargo, Citibank, KPMG, and IBM.
Many life-impacting decisions are guided by the same biases embedded in historical data, which AI systems could further entrench if left unchecked. For example, decisions about borrowing, employment, or medical treatment can shape an individual’s opportunities, financial stability, and quality of life for years to come. If AI algorithms are trained on biased data, they risk making these decisions in ways that systematically disadvantage certain groups, deepening inequalities instead of addressing them.
To combat this, it is crucial to recognize how biases in data can affect long-term decisions and work towards creating AI systems that account for and mitigate these biases. The next graphic, discussed in the Unequal Roots, Unequal Outcomes article, shows the model demonstrating the potential for high-impact bias in industries or activities with long timeframes for performance and significant risks. The following examples build on this model by illustrating typical decisions that could impact a participant’s life more than five years into the future, highlighting the potential for AI to either reinforce or challenge long-standing inequities if these systems are thoughtfully developed and managed.
Examples of Life-Impacting Decisions:
Bank Borrowing Decisions:
Description: Choosing to take out a mortgage, personal loan, or student loan can significantly impact a person's financial future, affecting their ability to save, invest, and manage expenses. The commitment to repay can last several decades, depending on the loan terms.
Potential Impact Period: Up to 30 years or more.
Employment Decisions:
Description: Accepting a job or choosing a career path can shape a person’s professional growth, income potential, and overall satisfaction. Staying with the same company or industry can influence one's life for decades.
Potential Impact Period: Up to 40 years or more.
Medical Decisions:
Description: Medical choices made early in life, such as lifestyle changes, surgeries, or treatments, can have long-term effects on an individual's health and quality of life. The consequences of these decisions can last throughout a person’s lifetime.
Potential Impact Period: Up to 60 years or more.
Education Decisions
Description: Choosing to pursue higher education, selecting a major, or attending a specific institution can have a significant influence on career opportunities, earning potential, and social networks. These decisions often shape a person’s career trajectory for decades.
Potential Impact Period: Up to 40 years or more.
Marriage and Family Decisions
Description: Decisions about marriage, starting a family, or choosing not to have children can affect the personal, emotional, and financial aspects of life. These choices often have lifelong implications, influencing living arrangements, financial planning, and social support networks.
Potential Impact Period: Up to 50 years or more.
Retirement Planning and other financial decisions
Description: Decisions about saving, investing, and preparing for retirement can determine financial stability and quality of life in later years. Early planning can help ensure comfort and security long after active working years have ended.
Potential Impact Period: Up to 30 years or more and beyond retirement.
Relocation Decisions
Description: Moving to a new city, state, or country can alter career opportunities, lifestyle, and social connections. The impact of relocating, especially during formative years, can shape long-term personal and professional outcomes.
Potential Impact Period: Up to 20 years or more.
Legal Decisions (e.g., signing contracts, wills)
Description: Signing legally binding documents, such as business agreements, prenuptial agreements, or wills, can have profound long-term consequences. These decisions often determine how assets are managed and distributed, affecting financial and personal well-being.
Potential Impact Period: Up to 30 years or more.
Predictive Policing and Sentencing
Description: The use of AI in predictive policing and judicial sentencing can influence how individuals are treated within the legal system, impacting criminal records, incarceration, and future opportunities. Decisions based on biased data can lead to systemic disparities that affect individuals' lives for years, potentially even a lifetime.
Potential Impact Period: Up to 50 years or more.
Conclusion:
The integration of AI into decision-making processes across various sectors brings both opportunities and risks. As highlighted in "Unequal Roots, Unequal Outcomes: The Deepening Bias in Modern AI and How to Uproot It," the potential for AI to perpetuate or even deepen existing biases remains a critical concern. Addressing this requires careful consideration of the data used to train these systems, along with proactive measures to ensure fairness. By doing so, we can leverage AI to make long-term decisions that promote equity and benefit society as a whole, rather than reinforcing historical inequalities.
Sources:
Hulett, Unequal Roots, Unequal Outcomes: The Deepening Bias in Modern AI and How to Uproot It, The Curiosity Vine, 2024
Friedman, M. & Schwartz, A., A Monetary History of the United States, 1867-1960, Princeton University Press, 1963
Thaler, R. & Sunstein, C., Nudge: Improving Decisions About Health, Wealth, and Happiness, Yale University Press, 2008
Kahneman, D., Thinking, Fast and Slow, Farrar, Straus and Giroux, 2011
Autor, D., Work of the Past, Work of the Future, Journal of Economic Perspectives, 2019
Cherlin, A. J. The Marriage-Go-Round: The State of Marriage and the Family in America Today, Knopf, 2009
Crawford, K., & Paglen, T. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence, Yale University Press, 2021
Comments