2021 BRN Discussion, page-14192

While not specifically about Brainchip, I feel this is worth mentioning because of the use of beneficial AI in the machine learning space and the recognition received. An Australian company Kablamo has been publicly recognised by Amazon for it's innovative bushfire prediction tool.
Australia is known for it's terrible bushfire seasons so it's also well overdue to have proper high tech tools to assist with managing them. It's ridiculous to think fires are still being managed by paper-based maps and manual processes in today's modern world when there is so much digital information easily available.

https://ia.acs.org.au/article/2021/australian-ai-firefighting-innovation-wins-big.html

International recognition has turned the work of “passionate” Australian data-analytics firm Kablamo into a global exemplar of AI innovation, after Amazon recognised its bushfire prediction tool as the most innovative AI/ML solution in its public-sector partner awards.

Since it was founded in 2017, Sydney-based Kablamo has built a team of nearly 50 developers, DevOps and DevSecOPs engineers, user-experience designers and agile managers.

It has built a reputation around its successful application of data analytics and AI/ML techniques to business problems in areas as diverse as home loan origination, automatic interview transcriptions, digital asset management, and more.

But it was work with the Victorian Department of Environment, Land, Water and Planning (DELWP) that led to the firm’s latest global recognition.

DELWP engaged Kablamo two years ago to help it explore the potential use of machine learning to better understand the spread of bushfires, whose almost randomness has seen them continuing to devastate Australian states despite a broad range of resources committed.

“We’re pretty good at building software that takes in pretty complex and large data sets, and good at user design and user experience,” co-CEO Angus Dorney told Information Age, “and we began looking at the way we could apply what we’re good at to the bushfire space.”

The problem soon became clear: despite having a wealth of historical data available, authorities have had no way of applying it in any meaningful way – ultimately managing fires using time-tested manual techniques and human intuition.

“We’re very outdated, in terms of technology and data, in the way that we’re managing bushfires,” he explained. “What is commonplace in large tech companies is so far off how our state and territory fire services are operating, in Australia and globally.”

This, despite a long history of damaging bushfires causing hundreds of deaths, burning millions of hectares and causing tens of billions of dollars in economic damage.

“There is still a lack of urgency and a lack of progress around how we use technology,” Dorney said, “especially at the top levels of government.”

As the project progressed, it became clear just how far behind authorities really were: “the more we looked into it,” Dorney explained, “the more we found that we are very reliant on things like paper-based maps, manual processes, and human-based decision making with very little data-led decision making on how we are managing bushfires.”

Yet useful data is already buried in a range of sources including satellite imagery, social media reports of bushfires, transcripts of emergency calls, cameras in national parks, sensors measuring moisture content of bushfire fuel in real time, location data of firefighting staff and assets, and even drone footage.

“The data is already there,” Dorney said. And while better utilising it “is a complex problem and not easy to solve”, he added, “you really have to ask whether we really have to be putting all those lives at risk by sending people in trucks with hoses to try and stop the mega fires, when there is all this other technology we could be using to help prevent them from getting to that place.”

NSW, for one, has leaned heavily on traditional firefighting methods in its response to the independent NSW Bushfire Inquiry, with authorities this month announcing $268.2m in additional funding, including $10.6m to support the new National Fire Danger Rating System and $5.2m for firefighting drones.

Pushing towards real-time analysis

The ultimate result of the project was a visualisation platform that, leveraging AI/ML algorithms trained on 40 years’ worth of historical data about bushfire movements and environmental conditions, has proven able to accurately model the risk of bushfires and their spread over time.

DEWLP is “able to generate predictive and risk analysis to a much higher resolution than was previously possible, and to ingest and analyse much more data than was previously possible,” Dorney explained.

“This has removed the technical bottlenecks that they were experiencing, as well as the limitations around the depth of the analysis that they were previously able to provide.”

Although the platform is not yet being used in real time, its design has been optimised for the ingestion of new data types and, potentially, its ultimate use as a tactical tool for fighting bushfires.