Rise of the intelligent machines

By

11 August 2017

A human-looking robot behind graphics of computer code

Artificial intelligence and machine learning probably conjure up images of robots taking over the world a la Terminator. Although machines rising up and becoming smarter than the humans that created them is not super likely (at least in the short-term), intelligent computers that can teach themselves how to solve problems is a reality in the world we live in now.

AI is already a pretty big part of our lives. Smartphones use AI for speech-recognition algorithms, and your Netflix is likely to show you the best recommendations based on your previous choices. And AI’s role is expected to grow to eventually become an everyday part of our lives.

So let’s first define some terms

What’s the difference between artificial intelligence and machine learning?

Artificial Intelligence (AI) is a computer system that exhibits some characteristics of human intelligence – which means it can complete tasks that humans need intelligence to do. Just like humans, intelligent computer systems need knowledge and data to decide how to act.

The benefit of AI is not that it can behave like humans can. Instead, it’s about using it to complete a task more efficiently.

Machine learning is used to make predictions and conclusions on the basis of data. Machines don’t “learn” as humans do – but they can be engineered to adapt to complex changing environments and adapt their responses based on the data they receive.

Animated image of Nintendo's Mario popping in and out of pipes

Mario’s not the only one investigating pipes.

Our AI and machine learning experts are writing programs that learn to improve themselves using vast amounts of data and in some cases, not much data at all.

From emergency detection to better management of a city’s infrastructure – we’re working with our partners to solve today’s problems, and build a safer and more efficient future for the next generation. Here are just a few of our favourite examples.

Pipe dreams a reality

Australian water utilities currently spend around $1.4 billion a year on repairs and maintenance to their infrastructure. When a pipe bursts, for example, there’s a lot of work and expense to shut it down, fix it and get it back online.

As we all know, prevention is better than cure. But to assess water pipes for leaks or vulnerabilities is expensive and disruptive – so much so that water utilities typically only inspect one per cent of their network assets each year.

That’s why we’re working with more than 30 utilities from around the world to develop data-driven analytics technology that accurately predicts pipe failure. Our technology helps them to better prioritise repairs, reduce operational costs of unexpected failure, and minimise the disruption to water supplies and the community.

Taking the high road

You’re sitting in traffic on your morning commute and you’ve missed the lights at the level crossing for the third time in a row – and you’re probably wondering if you’ll ever get to work on time. It’s pretty unlikely that you’re thinking about how your journey could be part of a bigger traffic picture. Luckily we are.

We are developing new machine learning techniques to create a range of tools to support safer, quicker and smarter roads.

By analysing and modelling traffic data, we can predict congestion, see hot spots for accidents and simulate how infrastructure changes would improve your daily commute.

Bright sparks save lives

Bushfires are dangerous to people and homes in their path – and even more dangerous to those brave enough to fight them. They’re also hard to predict. No-one can be sure where or when they will start, although well-educated guesses can be made. Once a bushfire has started it is also difficult to predict precisely where it will go.

No more rage against the vending machine required. We’re here to help!

So our clever researchers developed a tool called ‘Spark’ to predict and analyse fires. It takes our current knowledge of fire behaviour and combines it with state-of-the-art simulation science to produce predictions, statistics and visualisations of bushfire spread.

Mending vending

Do you ever wonder why the vending machine at the train station never has enough of your favourite soft drink or bag of chips? Why can’t they determine demand and plan the stock accordingly? Well now they can.

Our cloud-based AI software, Hivery, takes retail data from vending machines, stores and analyses it over a period of time and then advises the optimum number of goods based on the location and use of the vending machine. Ta dah!

Save

Save

Save

Save

Save

Save

Save

Save

Save

Save

Save

CSIRO's Data61

Find out more about how we’re using AI and machine learning to solve problems