The importance of diversity within AI modelling

Diversity is necessary in every organisation and it is also essential when creating AI system and algorithms.

Saba Samiei, founder at MACSO and winner of the Women in AI APAC Innovator of the Year Award for 2023 discussed with Digital Nation about the importance of diversity when building AI systems.

She said diversity in general is a must, whether its language, cultural, sexual, ability or talent diversity. And all of these factors impact an AI model.

“It has so many different angles that you need to think about because when you're developing a product that you're going to put on the world,” Samiei explained.

“Especially when you're doing applied AI, which is targeting a real world problem, it is impossible to do something, if you just have that one angle and nothing else, you will fail, that is your recipe for failure.”

Samiei said organisations need different people to sit at the table from those who develop the product to the ones who will interact with it.

She explained that when developing her AI-based sensors that detect sickness in livestock, having different opinions was crucial.

“When we were developing a product, we have mathematicians, physicists, AI engineers and computer scientists, but we also worked with farmers, vets, stocks people to understand how this whole thing is going to shape together,” she said.

For those organisations who are ready to take the plunge and have diverse staff building AI systems, Samiei said diversity brings disruption.

“Make sure that everybody is ready and accepts that disruption because it's no point hiring a diverse group of people just to get tick boxes on a piece of paper,” she explained.

“You need to listen to them, and you need to be ready to change with that new perspective that they're offering you. Otherwise, you're wasting your time, you're wasting their time, and you are going the opposite way.”

Digital Nation Australia is a proud media partner of the Women in AI APAC awards.