How can we use Gen AI to help Western Australians stay safe?
A hackathon with a purpose
idoba is proud to be Technical Partner of The Emergency WA Hackathon 2023, three days of creativity and innovation applied to a simple, but essential, challenge – Using Generative AI to build solutions that help Western Australians stay safe.
Gen AI, a force for good
We’ve all read the media hype and horror about generative AI – if humanity won’t soon be extinct, it will certainly be out of a job!
“In reality, the past year has seen the convergence of two things: the commoditisation of compute power necessary to run these LLMs and other neural networks, and an explosion of user experience and prompt engineering,” says Ed Lewis, idoba’s Director – Data & Analytics.
“Together these have driven a wave of innovation and applications built on technology that, in reality, has been bubbling away within business for some time.”
In pharmaceuticals, Amgen and others are using Gen AI to speed the development of new medicines. In manufacturing, Autodesk uses it to design better products and components more efficiently. Many businesses, like Adobe, are using the technology to simplify tedious, repetitive work and to support greater human creativity. Across business and society, Gen AI is already managing financial risk, spotting unseen medical conditions and making chatbots smarter.
At idoba we’ve built our own semantic language model, specifically for use in mining, to run domain specific queries against unstructured data like maintenance reports or emergency response plans, where documents are largely language-based and often handwritten. We’re also working on a voice-driven, virtual-assistant-like solution that can report technical and operational information, choose and run simulations and model business outcomes. Solutions like these support and accelerate better decision-making.
Co-creating a better future for clients using AI
Advanced AI (along with fantastic, talented people) is at the heart of idoba and we drive enormous value for clients through our Dynamic Driver Modelling and Gemini products.
These can create exact models and digital twins of mining operations that spot variations in real-time data. This enables engineers and managers to quickly take appropriate action. That could be making minute adjustments to re-optimise a process or taking timely action to avert an emergency. That's just a small step from allowing the AI to generate and instigate the appropriate action by itself.
One of Gemini’s use-cases is to create an exact digital twin of a mining operation, right down to every interaction of every truck and digger. With that, you can do two things. You can run very accurate simulations to understand what will happen if, for example, you change truck routing or even convert the entire fleet to electric trucks.
Or, secondly, you can monitor operations in real time, based on data from a myriad sensors, and respond to events immediately, before they become serious issues.
Applying AI on a territory scale
Imagine doing that on a Western Australia scale. Having the ability to predict natural disasters – whether storm, flood, fire or drought – and to simulate those disasters in order to develop efficient responses.
Imagine monitoring a wildfire as it burns. The wind changes direction and AI updates escape routes on the fly … and sends the information directly to car satnavs. But your secondary escape routes are getting too busy, so AI evaluates tertiary routes and directs people to safety by the optimal route.
Imagine real-time data and massively powerful AI spotting tiny changes in conditions and enabling firefighters to get ahead of the fire.
“There is, of course, a challenge,” Ed explains.
“Even in our current and growing climate crisis, wildfires are statistically rare. Do we have the volume of data required to properly train a Gen AI solution? How do prevent a Gen AI tool from hallucinating a wholly inappropriate solution?”