COVER STORY: The Women in AI winners creating change through the power of AI

Artificial intelligence and machine learning may be mainly known for its large language models and business use cases but in the right hands, it can impact and change every industry.

The Women in AI (WAI) APAC awards highlight the phenomenal work women are doing with AI in a wide range of industries.

Each of this year’s 18 awards categories are judged on innovation, leadership and inspiration; their global potential and impact, and the ability of their AI solution to provide positive outcomes for their community and citizens.

Digital Nation spoke to the main winners from this year to find out how they are revolutionising their industries through the power of artificial intelligence.

AI-powered sensors

Winner of the top gong of the night, the Innovator of the Year, Saba Samiei, founder of MACSO explained how she is using AI to make sensors intelligent.

MACSO aims to give AI the ability to replicate human senses to prevent catastrophes from happening.

Samiei said her and her team at MACSO make sensors intelligent for a variety of environmental monitoring mainly for early risk detection and prevention of catastrophes.

“Recently we have realised the platform be used for even more than that like ecosystem monitoring and basically different use cases as well.”

One of the first use cases was an intelligent audio sensor being used for respiratory health monitoring of pigs on pig farms. Samiei said it is similar to a human ear that listens to the pigs 24/7 and alerts the farmers as soon as it identifies a sick herd.  

The second use case is an aerosol sensor, which is currently being used for vape and smoke detection, that can be placed in schools, hospitals and hotels to monitor for dangerous chemicals in the air.

“It could be used in mining, in construction, basically any indoor air quality monitoring use case,” Samiei said.

Going forward, Samiei explained that she wants to bring all types of sensors into the platform.

“That could include vision, touch sensors, such as temperature, pressure gyroscopes, smell sensors and basically any other sensor that replicates what we do and sense.”

Winning the Innovator of the Year award said this award was a nice reminder for her to step back and take a look at all the hard work she has done.

“This award in particular for me because it's something that's outside of New Zealand, it was at an Asia Pacific level and it was judged by people I don't know and who don't know me. It was purely judged on the work that I have done,” she said.

“It means a lot, it's a validation. There are other people in the world that are looking at [my work], they're looking at everything I've done and they also believe in what I believe. On that front, it was a great feeling.”

Samiei said winning awards like the Innovator of the Year gives her a platform to showcase not only her work but her peers.

“Platforms like this, whether it be awards, grants, any sort of media, they are a platform that gives you a voice. I always aspire to be that voice on that platform to use it as a voice of those who don't have it,” she added. 

Preserving the Great Barrier Reef

Machine learning models and data sets are helping preserve the Great Barrier Reef.

First runner-up winner at this year’s WAI awards, Petra Kuhnert, group leader statistical machine learning group at Data 61 explains how they’re using AI to investigate pollutants within the Barrier Reef.

Kuhnert said working in the Great Barrier Reef preservation space is quite a difficult and political space to be in because you've got different kinds of methods that people use.

“I've spent a bit of time coming up with analytics and some data-driven AI techniques for quantifying pollutants and their loads into the reef,” she explained.

“[I’m] trying to understand for different catchments what the biggest sources of those pollutants are, and helping the Great Barrier Reef Foundation populate their Great Barrier Reef report cards.”

While her methods don't populate them now because they now have a source catchment model, Kuhnert said the very first report method wanted uncertainty in there to help understand how confident they were in quantifying these loads and where might they go out and do some further monitoring.

“As things have progressed, they've been using some of their models to do the reporting,” Kuhnert explained.

“But they always come back to my methods when things go pear-shaped, they can't run their model, or there's a gap in the data and we have to do some machine learning methods to do some interpolation and then bring it through my model,” she added.

Currently, she has moved away from Great Barrier Reef for now and is working in a number of strategic areas looking into the idea of emulating models.

“You might have a model of a bushfire, how a bushfire spreads or a model of crop production. These are tools that people have developed, they're deterministic, they don't quantify uncertainty,” she said.

“How hydrology works, how groundwater flow and sometimes these models are incredibly slow to run and you want near real-time estimates, particularly with a fire front moving and all sorts of things happening at the one time.”

Kuhnert and her team use machine learning to help speed up processes.

“These network models coupled with statistics or statistical thinking to speed up those models and provide faster solutions, more explainable solutions, so they can be run near real-time to make decisions,” she ended.

Early detection of cancer           

AI could be used in the future to help doctors and scientists detect cancer early on.  

Fatemeh Vafaee, deputy director at UNSW Data Science Hub and co-founder of OmniOmics and the second runner-up at the WAI awards, explains how a simple blood test could help doctors detect cancer early.

“We are working on is on a simple solution using a simple blood test, which is minimally invasive, and using that to diagnose cancer early on, as well as multiple other cancer management steps,” she explained.

Vafaee said due to the high dimensionality of data and the complexity of the underlying patterns within this data that are associated with the disease outcome, it's not possible to accurately find that sort of relationship between these molecular measurements or biomarkers and disease outcome without the use of AI.

“AI will help us to, at the first extent identify the important information within each modality that we have within each measurement,” she said.

“We measure from a simple blood or a few drops of blood, and then combine them effectively in the next level those set of heterogeneous information into a single predictive power predictive model.”

Beyond that, Vafaee said she wants to have a clear understanding of how the AI came to the decision to diagnose or detect cancer.

“We use explainable AI techniques to be able to look back into each individual patient and their molecular measurements and explain why that patient, for instance, has been diagnosed to have cancer or not. What are the contributing factors and how they are interrelated?” she explained.

“We think that this is of utmost important for both patients and clinicians to be able to know the source of the decision by such complex AI systems.”

Responsible AI

With the surge in the use of large language models, like ChatGPT, and more organisations looking to implement AI, questions around responsible AI continue to grow.

Winner of the Trailblazer of the Year awards, Dr Qinghua Lu, principal research scientist at Data 61 works behind the scenes to develop responsible AI patents for engineers who develop AI systems.

“My personal research focus is on responsible AI patent catalogue. In that patent catalogue, the current version includes more than 60 patents,” she said.

“The patents are reusable solutions for different levels, different types of AI stakeholders to use to make sure the development process and also the product design of AI system are responsible.”

In addition to the patent catalogue, Dr Qinghua and her team work on a question bank.

“The question bank includes over 300 questions for the AI stakeholder to understand or assess AI risk,” she said.

Recently, Data 61 has started new work on AI risk metrics.

“By doing that, we would like to help the industry to have concrete AI risk assessment metrics and environments,” she said.

“Instead of giving a subjective view on the AI risk, we want to provide a method for people to have evidence to support their AI risk results.”

Improving agriculture within African countries

AI systems are being used in Africa to understand the agricultural system within African countries.

Angella Ndaka, co-founder at the Centre for Africa Epistemic Justice received the Cultural Leadership in AI award for her work in agriculture.

She is working on how AI may impact the environment and how people that are co-designing AI systems for agriculture can bring voices from non-technical people.

“Especially the farmers and other actors in the farms, the people in the industry, and the policy people so the technology can be representing the voices, the values and the needs of the people that are in the farm," she said.

"Ideally, the technology is going to be used on the farm, it's not going to be used in the labs where it's designed."

At the Centre for Africa Epistemic Justice they are looking at the voice of the informal female farmer in an African context and how it's going to be represented by the AI systems that are being introduced in the value chain of agrifood systems.

“We have realised that many companies have started re-strategising, they've introduced AI to rate the people who supply food to them. Most of the female informal farmers have no access to internet, have no access to smartphones, so they'll not have access to any data-driven apps,” she said.

Informal farmers are those in developing countries that product goods such as cocoa, coffee and cotton on smaller plots of land with less resources than commercial-scale farmers.

Ndaka said the lack of internet to these female farmers is a point of exclusion which means their perception and perspectives may never be represented in that space.

“They also are likely to be closed out of business and out of work because that is their livelihood,” she said.

“We are looking at how do we get these women on board so they can also be represented, their perspectives can be heard, and we not going just to close them out simply because they informal. How do we integrate their perspectives in this?”

Digital Nation is a proud media partner of the Women in AI Awards. Read the full list of 2023 winners here.