What are power shadows?
A working definition from MIT Sloan
power shadows (noun)
Biases or systemic exclusion of a society, as reflected in data sets.
As a graduate student working on facial analysis software, Joy Buolamwini, SM ’17, PhD ’22, encountered a problem: The software did not detect her dark skin, though it easily identified the faces of people with lighter skin. She had to wear a white mask to be detected by the computer.
The problem was the power shadows that are cast in datasets, Buolamwini, a computer scientist, writes in her book “Unmasking AI.” The dataset she used — created by a government agency with the intent of representing a diverse population — featured images of public figures, often elected officials. Around the world, white men have historically held political power, and that was reflected in the data. The software she used was trained on images in which people were predominantly white and male.
Power shadows have serious consequences. Machine learning models used to inform hiring decisions could reflect past biased decisions made by humans. Models used to diagnose medical conditions could be based on data from mostly white patients, resulting in poor outcomes for Black patients.
The first step to overcoming power shadows is being aware of them, according to Buolamwini. “We must also be intentional in our approach to developing technology that relies on data,” she writes.
Working Definitions: Artificial Intelligence
MIT Sloan's Working Definitions explore the words and phrases behind emerging management ideas.
AI Executive Academy
In person at MIT Sloan
Register Now
These are the most urgent AI risks, according to 272 experts
Which AI risks could cause the most harm in the next five years? New MIT research shows that businesses should be aware of threats like competitive pressure and dangerous AI capabilities.
5 ways to make agentic AI a competitive advantage
Enterprises looking to make the most of agentic AI will have to rethink how work gets done and how teams are organized, without forgetting the human workers who set their companies apart.
Who will own the AI agent economy?
Here’s what businesses need to know as AI agents move from centralized systems toward a decentralized network of trillions of personal and organizational agents.
5 investments to close the gap between AI wealth and welfare
Transformative technologies like artificial intelligence succeed when societies make parallel social investments to ensure gains are distributed equitably, MIT Sloan researchers find.
Meet the new faculty members joining MIT Sloan in 2026
Debiased machine learning, the currency of invoicing, and training good models with bad data: Meet the new experts bringing their knowledge and skill sets to the MIT Sloan School of Management.
Data liquidity leads to AI success
Three levers — data architecture, data preparation, and data permissions — determine whether data becomes a reusable strategic asset or stays trapped in silos.
Pro-worker AI, explained
Artificial intelligence can make workers more capable and productive, but only if leaders design and deploy it to augment human judgment.
5 things to consider when working with AI
Researchers at the MIT Initiative on the Digital Economy share the latest insights about getting the most from working with AI, such as personality pairing and reorganizing job tasks.
Balance AI innovation and risk with ‘minimum viable governance’
As organizations scale generative AI, traditional governance models prove to be too rigid or too loose. Minimum viable governance calibrates oversight to risk, enabling responsible innovation.
The surprising power of warmth in AI negotiations
In MIT’s international AI Negotiation Competition, “warmer” agents achieved better outcomes in negotiations with other AI agents.