Certain elements of the media seem to enjoy scaring their readers with tales of a dystopian future where machines rule the earth. The problem with this narrative is that it’s humans that program the machines. It’s time to take responsibility for our actions and accept that it’s not computers that will determine our fate, but humanity.
Equally, when we look at artificial intelligence and blame it for increasing bias, maybe we should take a closer look at ourselves. Humans create bias and then program it into AI and machine learning algorithms. Perhaps the most unsettling fact of all is that technology merely imitates and enhances human decision-making and we are left with a black mirror that amplifies human prejudices.
Microsoft learned this lesson the hard way in 2016 when online users turned an innocent chatbot into a racist within a few hours. A few years ago, nobody stopped to question the ramifications of completing a harmless-looking Facebook quiz where your preferences would be sold on to the highest bidder. The information that was then fed into algorithms could target you with more posts you might ‘like’ to keep you on the platform.
The outcome of giving people more of what they want led users to set up camp in a private echo chamber where their opinions were enforced rather than challenged. But, as unscrupulous companies acted quickly to exploit their users through technology, why didn’t authorities act sooner to tackle the problem? and what have we learned?
Recent news that Facebook and YouTube have come under fire for providing offensive search suggestions to its users suggests that we still have a lot of progress to make. But we are becoming increasingly aware that the posts and pages that we like could be added to an algorithm that will deliver similar content back to us.
Surrounding yourself with content and people that do not challenge you will harm your growth and development. You don’t need a tin foil hat to question if this was all part of some sinister master plan that got out of control. These are just a few reasons why many are bravely stepping out of their echo chambers and the tribalistic nature of human thinking to embrace diverse thinking.
The need for diversity of thought is not restricted to technology. If your organisation is not as diverse as your customer base, how can it possibly serve your audience? An environment where everyone thinks the same is probably holding back innovation because the secret ingredient for creative thinking is diversity too.
Whether it’s online or offline, a diverse business strategy is becoming increasingly important. Rather than tweaking algorithms, we need to look closer to home at human nature itself. There is an increasing argument that the only way to tackle AI bias is through a diversity of thought.
Even the word diversity will mean something entirely different for each person reading this. Whether it be gender, race, sexuality, disability, religion or socioeconomic backgrounds, we are still only scratching the surface. Equally an individual’s age or social class could be preventing them from progressing forward in their career.
The only reason that we are beginning to question homogeneous thinking is that technology is magnifying our prejudices, and we don’t like what we are seeing. Could this digital wake-up call and lesson in self-awareness be a sign of progress? It might have taken a few centuries to recognise the bias against left-handers when it came to scissors, but we are improving.
If a combination of technologies such as AI, ML, and deep learning can help determine what we would like to watch on Netflix or know what our next Amazon Prime purchase will be, maybe we could put it to better use and help make a difference too. A tech startup called HiringSolved is an excellent example of how to use artificial intelligence to automate candidate matching, increase diversity and reduce time to fill jobs using RAI.
Companies such as Staples and Luxottica are already using RAI (Recruiting Artificial Intelligence, pronounced “Ray”). The tool makes it easier for recruiters to verbally set parameters to identify candidates, whether it’s by gender, skill, job experience or even veteran status. The technology also proved that by merely removing gender-based language from job ads, they were able to attract more female candidates.
For too long we have just accepted the way things have always been done, but we are beginning to questions everything and how every decision affects the world around us. Once again, this technology does rely on recruiters to set those parameters, but at least we are starting to recognise how important removing bias from AI really is.
Businesses are embracing the benefits of AI and machine learning because they can optimise it to complete a task efficiently. The problem is that nobody stopped to question if they should also tweak the algorithm to be optimised for fairness too.
With great power, comes great responsibility. If we are serious about removing bias out of AI, maybe it’s time to take a step back and take a long hard look at ourselves and our business processes rather than blaming technology for the problems that we have unwittingly created.