How It Works
The first step in utilizing and changing AI is understanding how AI works. The best way to summarize how AI works is that it is essentially a computer algorithm that is typically used to filter people by categories and requirements input by humans. The more information input into the AI systems, more refined results are generated.
How to Avoid Bias
You may have heard or read articles on recent issues AI has caused across industries. Most notoriously are law enforcement agencies, who have been battling the AI bias for quite some time. Recently, a report of an AI program used by a U.S. court to assess risk was biased toward the African American population – flagging black offenders twice as likely to reoffend than white offenders.1
Other lawsuits regarding bias using AI technology have occurred. In 2018, Amazon implemented past employee resumes as an outline for what they’re looking for in a candidate, but backfired as the AI discriminated against female candidates.2 In 2015, Google created an image recognition program that corresponded black people as gorillas.3 In 2016, LinkedIn created an advertising program that resulted in a bias against female names in searches.4 Also in 2016, Microsoft created a chatbot that was to learn Twitter formulas, but resulted in creating antisemitic messages.5 All of these lawsuits and issues are prime examples of when AI technically does exactly what it is told, but works against us.
The best way to avoid bias is to first understand how the AI will read the information. Understanding AI’s processing of information can help you determine what, how much, and words to avoid when inputting information into the system. Next, you might want to identify potential ways AI can misinterpret the information. This can aid you in forward-thinking solutions for problems the technology might cause. To learn more on AI, watch Janelle Shane’s Ted Talk on how AI breaks down the information humans input and ways to combat bias.
AI is still on the rise and implemented across industries. However, more state regulations have emerged to help regulate any potential discrimination or bias issues involved with its use. To ensure you are in compliance with local, state, and federal laws, please contact Harbor America. We understand how complex regulatory compliance can be, which is why Harbor America can provide you with a solution tailored to your compliance needs.
1 Machine Bias
2 Amazon Scraps Secret AI Recruiting Tool that Showed Bias Against Women
3 Google Says Sorry for Racist Auto-Tag in Photo App
4 How LinkedIn’s Search Engine May Reflect a Gender Bias
Source: Rise of the Racist Robots: How AI Is Learning All Our Work Impulses