Will 2021 be the year of ethical and explainable AI? Since I founded Omina Technologies in 2016 companies have a better understanding of AI but are struggling to get AI in production. I believe ethical and explainable AI can be the solution to get AI in production.

Since 2016 I have been giving speeches about the importance of ethical and explainable artificial intelligence (AI) in Europe, Africa and the US. Until recently, very few people outside of academia were talking about the importance of ethical AI and explainable AI.

In 2016 and 2017 companies still needed help understanding the capabilities of artificial intelligence. Managers at many companies did not understand the technology and had unrealistic expectations regarding AI. I often spoke with people who were scared of AI because they thought it would supplant human decision making.  Conversely, they thought AI was the answer to everything and would quickly and easily solve all their problems. Still other companies thought AI was only for the big players like Alibaba, Google and Facebook and questioned whether they too could leverage this exciting technology.

Five years later, at the start of 2021, companies have a better understanding of AI, yet according to Forbes, less than 15% of companies are actually using AI in production.  While AI is a game-changer, there is significant frustration with the technology. The Coronavirus has exacerbated the situation. A recent article in the Belgian newspaper, De Morgen, had the headline, “Corona has taught us we’re nowhere when it comes to artificial intelligence”.  But is this true? Why haven’t companies made bigger strides when it comes to AI? Why aren’t more companies using AI in production?

One answer to these questions is that companies are just starting to understand that AI is not plug and play.  If companies want to leverage AI in practice, they must ensure that algorithms are trusted, explainable and won’t create any liability issues down the road.  In other words, maximizing the benefits of AI also requires consideration and mitigation of risks introduced in adoption.

• The dangers of using biased data with AI are well known.  But did you know it is possible to proactively assess data for bias and mitigate many sources of bias in advance?
•  It is well known that predictions and decisions that cannot be explained to stakeholders or customers can be a significant problem.  But did you know it is possible to put more emphasis on explainability than performance optimization in AI?
•  It is clear that liability issues can arise when algorithms are used to make decisions that are unfair or biased.  But did you know it is possible in many cases to assess the potential liability of an algorithm before implementing it in production?

At Omina Technologies we have developed a powerful AI product, Justifai, that enables business users to put AI into production that can be justified to all stakeholders. Justifai allows business users to look for biased data proactively, it allows business users to specify the importance of explainability in predictions, and it helps business users to assess if they have created an algorithm that might be unfair or biased and, therefore, expose the business to liability or undue risk.  

Organizations must consider using AI technologies or risk being left behind by competitors.  But AI can only bring real value if it is trusted by all stakeholders.  The availability of AI that is ethical and explainable is a cornerstone to building that trust and Omina’s Justifai product is the tool to help companies achieve that goal.

Let 2021 be the year your company adopts ethical and explainable AI!

Rachel Alexander