IRVINE, Calif. (PRWEB) July 06, 2020
AI is gaining adoption fast. COVID-19 has accelerated advances in AI by many years.
While there are many benefits of AI, there are some aspects of AI that deserve closer scrutiny. One such aspect is the potential bias in AI models.
Why are AI models potentially biased? AI models rely heavily on historical data to engineer or predict a future course of action. If there is bias in historical data, it is likely that AI models will inherit that bias.
The AI Now Institute classifies such Bias related risks into two broad categories:
(1) Harms of allocation: This occurs when limited resources have to be distributed among individuals or groups. Job opportunities is an example. If you use AI to make hiring decisions based on actual historical hiring data, there maybe Gender bias.
(2) Harms of representation: This occurs when certain groups or individuals are treated unfavorably. A classic example is who gets selected for random screening at airports - potentially triggering Racial Bias.
Those of us who are responsible for shaping the future of AI have a huge role to play in eliminating any potential bias.
There are 3 urgent actions that can be taken to address this:
(1) Educate AI Product companies and start-ups in the area of Bias, especially unconscious bias that is not immediately evident.
(2) Hold companies and agencies that use AI for decision making accountable to ensure they perform due diligence before adopting an AI-centric approach to decision-making.
(3) Introduce legislation that will guarantee swift action if any Bias is proven in AI driven decision making.
To summarize, COVID-19 has accelerated AI innovation, and adoption. Once the models are in wide use, it will be difficult to course-correct. The time is now to advocate for #ZeroBiasAI.
What additional concerns do you have about potential Bias in AI? How are you going to contribute to #ZeroBiasAI?
About the author:
Pramod Kunju (“Data Guru”) is the Founder and CEO of Nakunj Inc, a niche professional services firm specializing in Data Analytics and AI. Mr. Kunju serves on the Board of National Diversity Coalition, an advocacy organization representing “Voice of the Voiceless”. Mr. Kunju is in a unique position to provide thought leadership on potential risks of AI adoption.
Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy or position of any organization.