24 May Artificial Intelligence Must Be Built On Human Diversity
One of the most important yet overlooked aspects of artificial intelligence (AI) is that it’s not intended to replace human brain power. In fact, current and future AI is best conceptualized as augmented intelligence. In this sense, the most successful AI supplements human intellect in a marriage of art and science, wherein the science component leverages current computing technology to facilitate the art of human decision making.
An easy way to imagine this concept is to consider the art and science of cooking. In this analogy, AI would be used to quickly retrieve, accurately measure and uniformly prepare ingredients, freeing the chef to focus on creating and executing their own artful recipe. Very few – if any – people would want to eat a dish entirely devised by a computer. Flavor is too nuanced and inextricable from human experience to imagine technology usurping the role of a fine chef in the foreseeable future.
If AI relies on human ability, then, just like the most popular restaurant will appeal to the most diners, successful applications must work for the overwhelming majority of people around the world. In this vein, employing diversity in AI and the tech industry as a whole isn’t just a virtue signal or a sociopolitical nicety; it’s an essential data-driven method for building the most viable product in an ever-growing sea of competitors.
An individual company’s most important measure of diversity depends on its particular product and target market. Often, these crucial diversity markers are hard to detect. So the process of diversifying should begin from the ground up; the more varied input you receive, the better you’ll be able to create the most consumable user experience with the highest quality results.
Diversity Goes Beyond Gender And Ethnicity
When YouTube released an early version of its video uploading app for iOS, it was surprised to see that up to 10% of its videos were uploaded upside-down. The whole design team scratched their heads and wondered what could have gone so wrong with their product. Eventually, they realized that the percentage of upside-down videos correlated with the percentage of people in the general population who are left-handed and tend to hold their phones 180 degrees differently than people who are right-handed.
YouTube’s mistake was assembling almost entirely right-handed teams to develop and test its app. The lesson to learn from this case is that, while women and ethnic minorities are most substantially underrepresented in tech – even for industry giants like Facebook and Google – it can be even easier to overlook other points of diversity that are less obvious but just as, if not more, important to the functionality of your product.