Artificial Intelligence (AI) is changing our lives at an unprecedented rate. Artificial Intelligence is being used to make decisions that directly affect millions of people. Most of the time people are not aware that Artificial Intelligence is being used. For the few people who know that Artificial Intelligence is being used to influence their lives, they have no way of knowing if biases have been removed from the Artificial Intelligence. People just have to trust the companies will do their best to reduce biases in Artificial Intelligence. Depending upon the motivations and the resources available in companies, eliminating biases in Artificial Intelligence might not be at the top of their list.
So, what is the problem? There is no central third-party authority to verify if companies have actually reduced biases in their Artificial Intelligence.
At its core, Artificial Intelligence is a set of instructions (aka algorithms) that use data to make decisions. There are three issues here:
1. People determine what the algorithms should do
2. People determine what data to use
3. People determine if they agree or disagree with the decisions made by Artificial Intelligence
As we can see, there are people involved in every aspect of creating and using Artificial Intelligence. Algorithms can intentionally or unintentionally create a disadvantage for people. In one example, algorithmic failures in facial recognition have resulted in people becoming unidentifiable due to their facial features. In another example, by not using real cancer patient data IBM’s Watson for Oncology made unsafe medical recommendations. These are just a few examples but they will continue to grow as AI becomes an essential part of our lives.
So, what can be done? I think we can solve these issues by creating a central authority/office at the Federal government level. The purpose of this central authority/office would be to inform the public if and where Artificial Intelligence is being used and if the Artificial Intelligence being used has any biases. Companies would inform this central authority/office about their uses of Artificial Intelligence and this central authority/office would certify if these Artificial Intelligence are bias-free. This central authority/office will have its own Artificial Intelligence to check other Artificial Intelligence created by companies.