Marc Andreessen famously observed, “Software is eating the world,” and according to Jamila Gordon, CEO and founder of Lumachain, so is Artificial Intelligence (AI).
AI and Machine Learning (ML) are well and truly ingrained in every industry. From healthcare, to agribusiness, to food manufacturing, AI is improving efficiencies and productivity, while also generating common challenges.
To better understand the use-cases and impediments of AI across different verticals we hosted a panel with three AI CEO’s who, coincidentally were all award winners in this years’ Women in AI Awards. They included;
- Dr. Michelle Perugini, co-founder and CEO of Presagen, a health care company building scalable AI across women’s health products
- Fiona Turner, co-founder and CEO of Bitwise Agronomy, an agtech scaleup that’s using AI to bring insights to farmers
- Jamila Gordon, CEO and founder of Lumachain, which uses AI to disrupt global protein and food supply chains.
Presagen’s first product, Life Whisperer uses cloud-based AI in the embryo selection process during IVF. The AI is trained using more than 20 thousand 2D embryo images to better identify the most viable embryos. According to Perugini, AI in women’s health products and in the fertility sector is not only bringing about efficiencies, but is also advancing accessibility to healthcare products and services.
“AI in our space, in the fertility sector, is really advancing patient outcomes. Its improving efficiency and standardisation within the clinic environment, its bringing global technology into clinics that wouldn’t otherwise be able to access or afford it. And it’s bringing affordable and accessible health care to patients around the world,” Perugini says.
Both Bitwise Agronomy and Lumachain are deploying computer vision, camera based AI, which works to mimic human vision. In agribusiness, Turner says that AI in the sector is “really starting to take off across all sorts of horticultural livestock and all facets of agriculture.”
Bitwise Agronomy claims that its AI solution delivers better results to farmers by using accurate data to improve yield and reduce costs.
Farmers can use GoPro cameras to capture footage during their work and upload this to the Bitwise Agronomy Platform. This then provides insights and uses historical data to make predictions around processes including crop performance, harvesting dates, climate impacts, water stress levels, sprays and irrigation systems.
Lumachain’s use of computer vision based AI is deployed to track the safety and security of the food manufacturing supply chain.
According to Gordon, Lumachain “provides an end to end set of modules for the global food supply chain, which shows where the food has come from, where it has
traveled, what were the conditions,” as well as ensuring that the products were “safely, humanely and efficiently produced,” while also ensuring the employees’ safety.
Challenges and Impediments
When it comes to the impediments of AI, the panellists across their varied industries, were in agreement that they are facing the same challenges.
According to Perugini, “I think there are some common challenges with respect to impediments to AI, mainly around data access and quality and quantity and type of data. And I think the world is kind of shifting their thinking around this. It used to be that everyone was trying to get the largest data sets. I don’t think it’s like that anymore.
“I think there’s a recognition that you need the right data sets. You need globally scalable data sets. Those data sets need to be representative broadly of the domain in which you’re using AI to solve a particular problem.”
When it comes to AI in healthcare, Perugini highlights the importance of broad data sets across multiple clinical environments with a wide range of patient demographics. Should these data sets not be wide enough, she says, then the AI will need retraining and rebuilding, leading to higher end-user costs.
“So everything that we do as a company is around solving that scalability challenge and getting the right data, which is globally diverse so that we can deliver these products
at scale and low cost,” she says.
In agribusiness, Turner talks to the same challenge using different language. She says, “It’s about how we curate our data sets.” This curation involves varied regions and growing types that are broad and deep enough to ensure that the AI is trained to multiple variables.
Ethics and AI
One of the key challenges that is facing AI globally is the rise of unethical AI. Rob Sibo, data and analytics Senior Director at the technology consulting firm Slalom Australia told Which-50 that the cognitive biases in the human thinking process are replicated in AI.
“Humans create the machine learning algorithms at the moment and a lot of times we propagate the same biases when we design the algorithms or when we collect the data that trains the algorithms,” says Sibo.
“There’s a lot of biases that get replicated into the machine learning models, which is what concerns me as well, because the model might be perfectly fine, the data might be fine, but the way we frame the problem and the objective is completely skewed. So you just apply a perfectly good model to a skewed problem.”
To mitigate the rise of unethical AI, The Australian Government’s Department of Industry, Science, Energy and Resources developed an AI Ethics Framework, which it claims helps to achieve better outcomes, reduce risk and encourage good governance.
According to Perugini, who helped to develop the framework, it is reminiscent of Australia’s strict regulatory framework for healthcare.
“I think it’s a very structured and strong way to manage risk around how many people are you going to impact with this AI, what is the outcomes of kind of getting it wrong and how do we therefore mitigate those risks or ensure that the right data has been utilised or that testing has been done to protect the consumers that we are serving,” she says.
The Future Of AI
“AI will be in every industry in some way,” says Gordon, and “AI will impact every aspect of our lives.”
AI is set to become even more integrated into our lives than it already is, and according to Turner, so much so that we won’t even know it is there.
Looking to the future, Gordon sees human collaboration with AI to be the next big step where AI can play a supervising role, as well as automating manual tasks is set to increase.
The integration between AI and robotics, according to Turner, will be one of the greatest drivers of efficiency where the AI can act as the “brain and the vision” to the robot’s physical counterpart.
“I think quantum computing is going to help accelerate the growth so we can eventually get to General AI, which is a fair way off where your AI can do multiple things at once,
“More these kind of futuristic AI robots that you hear of. I think we’ll get there, but we’re a fair way off from that.”