We Can Fix the AI Bias Problem

As we enter a new decade, we have been pondering the rise of the algorithm – those secret formulas able to accurately predict medical diagnosis or recommend what to watch next on Netflix. With so many personal and professional experiences guided by algorithms, the role of bias is a global problem with several local solutions. 

The question facing us as we begin a new decade is what, if anything, can be done about biased algorithms. As one of the global leaders in AI and governance, the UAE is positioned to take the lead on fixing the problem of bias. Home to more than 200 nationalities and boasting one of the best knowledge economies in the region, the country is suited for this globally important task. 

Wetware is hard to update

But where do we start? According to Sendhil Mullainathan, a professor of behavioural and computational science at the University of Chicago, the problem of bias isn’t actually as bad we might think. Writing in the New York Times, Professor Mullainathan argues that “changing algorithms is easier than changing people: software on computers can be updated; the “wetware” in our brains has so far proven much less pliable.”

Algorithms merely highlight the prejudices and preconceptions of those who create them. The algorithm, in and of itself, is little more than a vehicle. One that can be changed and updated with ease. The challenge is figuring out how to make those changes and here we have to get creative. 

For Professor Mullainathan, the solutions lie with sensible regulations. In 2018, he “took a cautiously optimistic perspective and argued that with proper regulation, algorithms can help to reduce discrimination. But the key phrase here is “proper regulation,” which we do not currently have. We must ensure all the necessary inputs to the algorithm, including the data used to test and create it, are carefully stored. Something quite similar is already required in financial markets, where copious records are preserved and reported while preserving the commercial secrecy of the firms involved. We will need a well-funded regulatory agency with highly trained auditors to process this data.”

Small states can have a big influence

Sensible regulations designed to protect individuals while ensuring that creators and businesses remain innovative are critical. This is especially true for issues such as data privacy but can easily be expanded to oversight over algorithms. Smaller countries with strong technology sectors can and should take the lead on these issues. 

Countries such as Estonia, Singapore, and the UAE have spent the last two decades investing sustainable amounts into their knowledge sectors and now punch far above their weight in the tech sphere. Given their size, they can more aggressively explore sensible regulations needed for these pressing issues than much larger countries such as the United States and China.

The UAE Ministry of Artificial Intelligence has already made headway on this issue. Omar Sultan Al Olama, Minister of State for Artificial Intelligence, confirmed in April 2019 that the UAE government adopted an integrated and dynamic model for the utilization of artificial intelligence that supports industry growth, development of new sectors as well as strengthening governance and ethics frameworks, ultimately anticipating future challenges and creating a positive change for humanity. Critical to this mission are resources devoted to countering bias in algorithms. Other entities such as Smart Dubai have adopted similar AI ethics codes that address the bias issue. 

Technology creation and regulation is shifting away from the West to cities in the emerging world like Dubai. The coders and developers in these new centres of innovation are exposed to a vast array of influences that aren’t present in places like Silicon Valley. By finding solutions to global problems like bias in algorithms, the UAE can further entrench its position as one of the most innovative countries in the world. The will, infrastructure, and people are all present. Now it’s time to solve these problems that affect us all. Algorithms, after all, are only as biased as those who write them.