Artificial intelligence (AI), as sci-fi as aliens, is now a fundamental part of our daily lives. AI is a field that leverages computer systems, statistics and mathematics to mimic how animals, mainly humans, solve problems and make decisions. AI can do amazing things, recommend the best movies for you on Netflix, find exactly what you want on Google, keep you on TikTok for hours and more recently talk to you so naturally that you think it’s another person.
However, for all of AI’s amazing feats, it has a not-so-hidden secret, Bias. The most prevalent forms of AI rely on data and human experiences to work, meaning any flaws from either of those; human biases, unreasonable assumptions or incorrect data can sneak into the AI system, and impact the decisions it makes. The impacts are often so ingrained that it’s hard to uncover them, unless you’re looking for them. A few notable instances are:
1) An AI powered image cropping tool that cropped out persons of colour and the elderly, because it was only exposed to young white persons and did not know how to handle darker skin tones.
2) A system used to automate the decision of persons getting bail vs jail time perpetuated racial biases, researchers noted it was built using data skewed by years of racial discrimination. You may be saying, “that’s stupid, what kind of mistake was that?”. I’ll tell you another secret, AI is obedient and not very smart. AI is very good at very specific tasks and learning rules from data, but if the fuel for the rules is incorrect it won’t know. All those movies where AI takes over the world and tries to eliminate humans, I blame the crazy person that built the AI, that person does not have many friends. But, what can we do to make AI fair for everyone? We make sure everyone is represented when building the AI; diversity and representation is the key.
Diversity is the condition of having or being composed of differing elements, applying this definition to persons, it means different races, genders, physical capabilities, cultures, ages, and so on need to be considered and consulted. As we seek to make life better for all humans we need the diversity of human intelligence as a blueprint for thinking machines. Jamaica’s own AI company, StarApple, utilizes this diversity as a superpower when building AI models, using teams composed of people with diversity of backgrounds, experiences and thoughts.
AI’s uses in daily activities are growing exponentially and impacting people from all backgrounds, yet the majority of the AI development and funding is concentrated in a few countries and driven by relatively uniform types of persons. With increased diversity and improved representation we can ensure AI is developed with the Caribbean in mind. Our value to the rest of the world is greater Innovation and Productivity.
Greater Innovation
To innovate means introducing changes and new ideas to the way something is done or made. Technological evolution is happening at an exponential rate. If outdated practices, beliefs and principles continue to rule we will find ourselves coming to a screeching halt. On the other hand, with an increase in diversity and representation, industries are incentivized to listen to all consumers. Diverse teams have varying thought patterns, experiences, and skills at their disposal that will bolster developments in AI and business in general.
Greater Productivity
Productivity slows when there is a skills deficit. AI’s development demands different skills, unlikely to be found in a single person or region. Even the simplest projects require different levels of skill in various areas.
Differences in beliefs, approaches and ideas don’t have to be faultlines from which conflict emerges, but opportunities for us to expand the AI industry. The Caribbean is the perfect development ground for AI, our diversity per mile and innovation per person are our largest natural resources. Join me in building a Caribbean AI Industry built on Diversity and Fairness.
Stay Innovative Jamaica.
Source: Jamaica Observer: https://bit.ly/3PTgKqV