Data privacy is a matter of civil rights and liberties
Illustration: Getty Images
Every day, our personal information is collected and used in ways we may not realize.
Some companies make money selling digital profiles that might include your social media activity, television choices or mobile app usage.
The challenge: “Problems can occur when algorithms are trained using pristine data, and then deployed with little or no realistic field testing and review,” explains Dr. Shirley Cavin, a Leidos AI expert. “Real-world data has anomalies and can result in algorithmic outcomes that are unintended or inaccurate.”
Dr. Cavin said that when an algorithm is given decision-making power, there should also be a human in the decision loop to help recognize algorithmic errors and correct them, at least until the algorithm is proven reliable.
AI technology development is also outpacing new laws to regulate it, according to Leidos Chief Privacy Officer Heidi Salow.
“Legislation can take years to catch up with technology like AI,” Salow said, “and new domestic and international privacy laws add to the complexity. We’re constantly refining our global data protection compliance program to address evolving legal requirements, data ethics and best practices.”
Leidos impact: Leidos scientists are developing solutions that give algorithms greater transparency and allow AI technology to meet high ethical standards.
For example, the company recently created the Framework for AI Resilience and Security (FAIRS), which detects and manages risks in AI applications, including algorithmic bias.
“We’re shaping trusted AI components through a human-centered approach in direct collaboration with the technology owners, technology users and the public,” said Roopa Vasan, a Leidos senior scientist on the project.
Meanwhile: Results from an independent study show that some companies, including Leidos, are committed to using technologies in a way that encourages privacy and eliminates bias often found in human analysis.
“I think it is important to educate and advocate for best practices, policies and laws that balance privacy, civil rights and civil liberties with the need to enhance security, efficiency and convenience,” said John Mears, Leidos Tech Fellow, Homeland Security Solutions.
Looking ahead: Leidos plans to launch an ethical AI governance board that can help assess AI-based projects and their impact on marginalized stakeholders.
“The board will include a diverse group who represent a cross section of ideas and experiences to reinforce equity and ensure transparency in the AI and machine learning lifecycle process,” said Beverly Thompson, a Leidos senior scientist.
January 28 is Data Privacy Day. You can use this guide to help protect your personal information.
Learn more about our company's commitment to data privacy.