Artificial Intelligence is being hailed as the brilliant technology that can help businesses increase their operational, predictive, and productive power. Organizations are willing to invest heavily in this disruptive technology that is transforming existing workflows. With AI being omnipresent – from smartphones to household finances to law and justice systems, it is extremely important that both the public and the companies realize that there could be a potential dark side to artificial intelligence too.

Many movies speculate on how artificial intelligence can escape man’s control to reign supreme, subjecting man to slavery. While this is a bit far-fetched, the actual danger is much more subtle.

In an article from the Live Mint, Sandipan Deb writes that AI is only as good as the data that is fed into it. “The data is worked on by deep-learning software, which absorbs the data, figures out patterns, creates rules to fit the patterns, and keeps tweaking those rules as more data is fed into it.” The data, which is fed by humans, will contain the prejudices that mankind holds, which will ultimately influence the end result that reflects the societal biases like racism, sexism etc.

In May 2018, a report highlighted that an AI-generated computer algorithm used by a US court for risk assessment was biased against black prisoners. The program asserted that blacks were twice as likely as whites to re-offend in the US. This conclusion was a result of the flawed or skewed training data that it was learning from. While machine learning is often hailed as being impartial and unbiased, the technology will only be as good as the data that has been fed into it.

In 2015, Google came under severe criticism when its photo app tagged two black people as gorillas—perhaps because the algorithm’s training data set did not have pictures of enough black people.

In 2016, Russian scientists ran a global beauty contest to be judged by an AI. Of the 44 winners, only one had dark skin. The algorithm had been trained mostly with photos of white people, and it had equated “fair skin” with “beauty”.

Another example is when an AI is fed the resumes of candidates for a top corporate job, and it chooses a man, because data shows that men have overwhelmingly outnumbered women as CEOs in the past. Going by the data, the AI will decide that a man will make a better CEO than a woman. While the woman would have been pushed back due to gender bias in the past, the computer would not have any idea about this as it is powered by the data it is given.

While AI is being increasingly deployed across a wide variety of domains, from personal digital assistants, email filtering, fraud prevention, voice and facial recognition and content classification to generating news and offering insights into how data centres can save energy, the discrimination that AI could implement should also receive attention. More than a technical issue, this remains a social problem that technologists could find difficult to solve.

This blog was written by our Content Writing Intern – Rona Sara George. Click on the name to view her LinkedIn profile.

Author: Xaltius (Rona Sara George)

This content is not for distribution. Any use of the content without intimation to its owner will be considered as violation.