How to Avoid Bias in Emerging Technologies and AI

Emerging technologies, including artificial intelligence (AI), machine learning (ML), quantum computing, and blockchain, hold immense potential to revolutionize society. However, their development and implementation come with the risk of embedding and perpetuating biases, particularly those against individuals with disabilities. To ensure the responsible evolution of these technologies, it is crucial to mitigate biases at every stage, from creating a Minimum Viable Product (MVP) to an ideal-state solution. Addressing bias effectively involves a holistic approach that considers not only the technical aspects but also the societal context in which these technologies operate.

The National Institute of Standards and Technology (NIST) emphasizes the importance of looking beyond machine learning to understand the broader societal influences on technology development. Their guidance highlights that biases can originate from algorithms, data, and the societal context of AI usage. This comprehensive perspective is critical for identifying and addressing the root causes of bias. According to the White House’s Office of Science and Technology Policy (OSTP), algorithmic discrimination can arise from various factors, including biased training data and flawed sampling methods. Such biases can unjustifiably disadvantage individuals based on race, color, ethnicity, gender, disability, and other protected classifications, perpetuating historical and social inequities.

To combat these biases, the OSTP recommends a set of principles aimed at promoting equitable design and usage of automated systems. Key actions include conducting equity assessments during the design phase, ensuring the use of representative data, prioritizing accessibility, and engaging in ongoing testing and mitigation efforts. Transparency and accountability are fundamental, necessitating clear reporting of algorithmic impact assessments and mitigation strategies. Additionally, embracing diversity and inclusion practices is essential to empower marginalized communities and counteract implicit biases that may exist even among executives and decision-makers. By focusing on objective criteria like error rates, task completion times, and privacy concerns, rather than subjective perceptions, stakeholders can make more informed and equitable decisions regarding the deployment of emerging technologies.