Bias in Artificial Intelligence: Curious Case of Google Translation

By | Et Alia, Featured

Google-AI-Bias-technology2


You may wonder why Google Translate is exhibiting such striking gender stereotypes. Haven’t we all seen women doctors or male gynaecologists around us? So, what’s the big deal? Why Google machines’ artificial intelligence can’t get it then? Can we blame the machines for such harakiri? Is it their fault?

No, it’s not. Machines are automatically learning algorithms. Algorithms, by the way, many believe are opinions expressed in codes. Are we developing biased algorithms then? Should developers and architects be wary of bias creeping into the systems? Aren’t machines simply improving their own experiences by picking up deeply ingrained race and gender prejudices concealed within the patterns of language use? Yes, it’s a reflection of our collective, deeply ingrained race and gender prejudices. If robots are racist and sexist, it is us who create them. If AI programs exhibit racial and gender biases, it echoes our social inequalities and prejudices which AI is learning.

 

Last week, Sam Levin, a reporter for The Guardian has written an in-depth article titled, Bias deep inside the code’: the problem with AI ‘ethics‘ in Silicon Valley‘ It talks about how algorithms play a growing role in criminal justice, education, tech advisory boards, and academic programs and how all these mirror real-world inequalities. Read it here

Meanwhile, Google in its efforts to “promote fairness and reduce bias in machine learning,” is working towards providing feminine and masculine translations for some gender-neutral words. The Telegraph writes,

In an interview with The Verge magazine last month, Google’s head of translation, Macduff Hughes talks about three big initiatives on the fairness and bias issue. Click here to read the interview.

Meanwhile, Google appoints top Oxford University philosopher to sit on a brand new ethics council that includes philosophers, engineers, and policy experts amid concerns surrounding the company’s use of artificial intelligence (AI), reports MIT Technology Review.

Google is trying its best to keep AI bias at bay. I will keep a hawk eye, because Google being Google, the first mover, sets the bar for all. Isn’t it?

 

Brands & Bias Discrimination by Algorithm

Source: 1. Amazon: Reuters  2. Microsoft: MIT Research (PDF)   3. Facebook: MIT Technology Reviewezy insights  4. Alexa, Watson, Einstein & Siri: Harvard Business Review  5. Bing: The Economic Times

How AI Systems Amplify Bias

Source: SearchEnterpriseAI, TechTarget

 

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *