Reshaping virtually every industry, profession and life, artificial intelligence (AI) has already had a profound impact on our future and will continue to do so. Natural language processing algorithms that convert speech to text are found your smart home applications, while predictive analytics that use deep learning to forecast behaviour can be found on every digital platform and device.
Fourteen of the world’s most advanced economies have announced a total of $86 billion in funding for AI programs and activity, while digital innovation has been slated to deliver $315 billion in gross economic value to Australia over the next decade. This illustrates just how critical an ingredient AI is in our ongoing economic success.
Artificial Intelligence (AI) is a broad term used to describe a collection of technologies able to solve problems and perform tasks without explicit human guidance. Some of these include: machine learning, computer vision, natural language processing, robotics and deep learning.
A general-purpose technology, AI uses data-driven algorithms to autonomously solve problems and perform tasks without human guidance. The algorithms that underpin artificial intelligence have existed for quite some time, however exponentially growing volumes of data and the widespread availability of affordable computation mean that Australia, and the world, can operate this revolutionary technology at a scale and speed never seen before.
AI represents a significant opportunity to boost productivity and improve the national economy through its strong potential to enable industry to make better products, deliver better services, faster, cheaper and safer.
As Australia’s national science agency, CSIRO’s work in artificial intelligence is 89 per cent more cited than the global average, according to a Normalised Citation Impact, InCites (Web of Science Articles & Reviews, 2014-2018).
AI is being used across all domains at CSIRO including: gene sequencing in crops to improve grains in response to changing environments and market needs. In our oceans, it is used to identify fish species to improve seafood provenance and provide more accurate data for sustainable fishing. In cities, it is used to predict the failure of infrastructure like the Harbour Bridge or water pipes and sewage systems. In hospitals, it helps to forecast demand to ensure access to emergency care.
CSIRO has invested $19M into an Artificial Intelligence and Machine Learning Future Science Platform, to target AI-driven solutions for areas including food security and quality, health and wellbeing, sustainable energy and resources, resilient and valuable environments, and Australian and regional security. We are uniquely placed to deliver national AI value because it is at the intersection of government, industry and the research community, leveraging the best talent in the country to tackle problems.
We are exploring how AI can accelerate the identification and isolation of genes in crops, enhancing our ability to improve grains and develop future applications that can respond to market needs.
For example, CSIRO has sequenced the gene variation in the OzWheat collection. Then, using machine-learning, new ways to combine this genome information with climate records and field performance data was developed.
This has changed the way researchers link variation in genes and gene pathways to crop performance, and has provided insights into how a century of selective breeding has adapted wheat to Australian conditions.
An online map-based tool that uses machine learning to collate and display spatial data from Australian government agencies, helping 543,000+ users make the most of high-quality and predictive data model of the country.
Over 10,00 datasets have been interwoven into a single platform.
We use structural health monitoring technology that employs AI and machine learning to collect and segment data analytics that help inform and enable maintenance, productivity and asset life extension.
We’re collaborating with 30 utilities from around the world to use machine learning and data-driven analytics to accurately predict pipe failure. Data is used from past failure events, including water pipe age, construction material, pressure and size to identify the likelihood of failure and prioritise repair or replacement.
Our Patient Admission and Prediction Tool (PAPT) is a new suite of software tools for hospitals, accurately forecasting demand and helping to ensure access to emergency care and a hospital bed, delivering 90 per cent accuracy.
Our researchers are currently using machine learning for the early detection of cancer. Autodensity is an algorithmic breast density measurement software package that is a fast, automated and practical tool that reliably measures breast density from mammograms that presents a more efficient than labour intensive breast measurement techniques used in research settings.
Spark is a toolkit for the end-to-end simulation and analysis of wildfires. Users can design custom fire scenarios through program modelling software, creating an accurate representation of future situations, an element that can greatly assist emergency management decision makers for predicting risk, deploying firefighting resources or planning evacuation routes.
Data61 researchers are monitoring biodiversity in the Amazon using wireless sensor networks and using machine learning to continuously track and measure changes in the environment.
Our researchers are combining advanced materials, sensing, and autonomous robotics technologies to evaluate the reef health, detect illegal fishing and shipping, monitor and manage threats.
CSIRO’s Data61 Mixed Reality Lab is a technology game-changer, digitising the full manufacturing value chain, enable real-time situational awareness and lead to better decision making and planning.
A piece of software currently used in the real estate industry, Accurait uses AI and machine learning in its text classification tools, optical character recognition (OCR) and a company’s metadata to drastically reduce the document-intensive process of commercial lease abstraction by at least 30 minutes per lease.
Our Risk Lab team uses machine learning to identify anomalous events in financial data that may indicate attempted fraud cases. It is also used in the analysis of hedge fund data using clustering methods to identify groupings to inform better decision making and better understand volatility.