Beware the Potential Implicit Bias That Comes With Artificial Intelligence
Companies are increasingly looking for ways to address the impact of unconscious bias—and for good reason. Unconscious bias can have negative impacts on inclusion and diversity efforts, as well as general decision-making. Artificial intelligence (AI) is seen by many as technology with the potential to take the human element—and bias—out of the equation entirely.
The problem, though, is that AI is created by humans. Consequently, AI, as well as the data it relies on to make decisions, can lead to insights and decisions that are still subject to bias. Worse, magnified by computing power, biased AI could end up being even more harmful than biased humans.
AI and Bias Creep
We don't have to rely on hypothesis and speculation to see the potential for bias to creep into AI. Several prominent companies have struggled with tasking AI to assist in human resource activities. "Efforts to use AI in recruiting have hit some rough spots," says Monica Melton in an article for Forbes. "Last year, an in-house project by Amazon to build a tool to select the top candidates using artificial intelligence was shelved after executives realized the program was penalizing resumes from women. The reason, according to Reuters: The AI system had taught itself, based on past Amazon hiring patterns, that the most successful candidates were male."
Melton gives another example from LinkedIn, which discovered that the recruiting algorithm it uses to highlight prospective candidates for recruiters was at risk of compounding hiring bias by, for example, highlighting more males than females.
Beware “Artificial Intelligence”
AI has the potential to solve a number of the problems facing our society as well as improve and increase the efficiency of many of our day-to-day activities. But "artificial intelligence" is a somewhat misleading term. AI is influenced by humans and shares many characteristics and flaws with humans. Assuming that such a tool will be entirely impartial is a mistake—AI is only as non-biased as the information that fuels it.
It's fortunate that the companies that have so far attempted to use AI to streamline its hiring processes were vigilant enough to identify potential issues on bias on take corrective action. We can learn from these lessons.
OUR NEW BOOK HAS BEEN RELEASED!
In Inclusion: STILL the Competitive Business Advantage, we continue our contributions to thought leadership on the importance of inclusion in an environment that has been roiled with new discussion—and new dissent—amid rapidly changing demographics, continually emerging technology and a global economy that is continually shifting to favor newly emerging market powerhouses. We're very gratified by the positive reviews already pouring in.