Ethics of A.I. 2019
Full unedited live stream recording is available here.
Artificial Intelligence (AI) is one of the most important topics of today. Developments in AI will disrupt the economy and the world in ways not seen since the industrial revolution. As AI continues to permeate every aspect of our lives, we must ask ourselves:
- How can we ensure AI is being used for good?
- Who is accountable for the decisions made by AI?
- What is the responsibility of government, corporations and the general public in shaping the rules of how AI will be created and used?
These are questions that need answering. The Artificial Intelligence Collaborative Network (AICN) will host an entertaining evening which will inform and educate people about the emerging debates in the field of the Ethics of Artificial Intelligence (AI). As the impact of AI increases on society it is important for people to be informed about the critical ethical decisions involved when creating applications using AI.
On August 15th, 2019, AICN brought together world-class experts in AI, technology, law, policy, media and ethics for thought-provoking presentations and a moderated panel session in Adelaide, hosted screenings at 6 other major cities and made available as live stream reaching over 700 people across Australia.
14:50 Yolanda Sam - Co-founder AI Collaborative Network (AICN)
17:55 Prof Caroline McMillen - Chief Scientist for South Australia
30:35 Ellen Broad - Data Expert, Author "Made by Humans: The AI Condition"
54:01 Dr Mor Vered - Human-Centred Explainable AI, Monash University
1:08:02 Dr Niels Wouters - Inventor "Biometric Mirror", Science Gallery & Melbourne Uni
1:12:08 Live Demo - Biometric Mirror
1:27:45 Panel Discussion facilitated by Dr Andy Stapleton, featuring Ellen Broad, Mor Vered, Daniel Kiley and Terry Sweeney
2:19:15 Yolanda Sam - Connecting with the AICN community
Predictive Maintenance: Why and How
By Latha Madhuri Pratti (Data Scientist, PREDICT Australia) and Amélie Froger (Australia Manager & Business Development, PREDICT Australia)
In today’s world, we have limited resources and the competition is serious. It is a real need for the industry to become smarter and use what already exists such as data, along with the engineering expertise in modelling the systems. What we do at Predict is to try to make the most of this data to perform Predictive Maintenance.
The solution developed includes health assessment, failures predictions, failures root causes diagnosis, throughput improvement and energy optimisation, therefore maximising systems performance, asset availability and reliability. By detecting and analysing subtle drifts from monitored signals, PREDICT solution can identify:
- What is going to break?
- When is it going to happen?
- How, where and why?
Digital Solutions to Mining Challenges
By Dr Larissa Statsenko
PRIF Consortium and Mining Strategic Research Manager, Institute for Mineral and Energy Resources
The mining industry is moving towards the digital transformation. The key to interoperable, agile and efficient operations, is a fully integrated mining value chain that enables rapid and well informed decision making, based on near-real time data streamed from advanced sensors and automated equipment. The Research Consortium led by the University of Adelaide has been established to provide digital solutions for mining value chain integration. Mining Industry 4.0 CRC, another Australia wide collaborative initiative, has recently been established to facilitate digital transformation of the mining industry.
Exploration of AI Possibilities
By Andrew Harris
Mining Analytics Consultant - Consilium Technology
Consilium Technology is discovering opportunities for companies to benefit from AI, and developing AI Tools that improve decision-making. AI brings huge benefits in today’s data-driven world. It’s simply not possible for people to make fast and accurate predictions, or optimal decisions, from the influx of large, complex and multi-dimensional data. AI can provide more accurate predictions, which reduces uncertainty, allowing better decision-making. Rather than fear being replaced by robots (substitutive AI) in the workplace, there’s an opportunity to embrace a human-machine partnership (complementary AI). Some case studies of AI in mining will be discussed, including process optimisation, crew scheduling and predictive maintenance.
Mining 4.0 - A Geologist’s Perspective with contributions from the world of IT
By Professor David Giles
Prof David Giles is Strand Leader and John Ralston Chair of Minerals and Resources Engineering at the Future Industries Institute, University of South Australia. At the Future Industries Institute we partner with resource sector companies to deliver innovative solutions underpinning productivity and sustainability across the resources industry value chain.
Prof Giles has over 20 years’ experience in minerals exploration spanning the boundaries of industry and academia. Between 2010 and 2018 Prof Giles was Program 3 Leader in the Deep Exploration Technologies CRC and part of a multidisciplinary R&D team responsible for introducing Coiled Tubing drilling with attendant in-field sampling and sensing technology to the mineral exploration industry.
He is Chief Scientific Officer of MinEx CRC where he manages a portfolio of mineral exploration research projects in collaboration with resource companies, METS companies, government organisations and research providers.
GamerAI: Building an AI assistant for esports teams and enthusiast players
By Dr Grant Osborne
Chief Technology Officer - Gamurs Pty Ltd
eSports (or professional gaming) is a rapidly growing industry, with tournament viewership already surpassing traditional sport broadcasting. By 2020, the eSports industry is estimated to be worth US$1.4 billion. As the industry becomes more competitive, GamerAI will equip eSports players and teams with insights and strategies to maximise their chances of winning every competitive match they play.
The GamerAI team are developing AI tools that can break down game footage into a log of events or actions which are then compared to match outcomes. The GamerAI team is working directly with esports teams to measure performance and validate this technology.
This talk will discuss the challenges faced by eSports teams, the technology being developed to solve these problems, and a demo of GamerAI in action.
Automating microscopy for better environmental monitoring
By Jordan Gruber
SA Young Achiever of the Year 2018, CEO/Founder Fronter Microscopy
Frontier Microscopy’s mission is to accelerate scientific discovery by equipping scientists with intuitive tools to simply automate any microscope analyses.
Scientists equipped with Frontier’s inexpensive microscope hardware will use their cloud-based automation software. This software allows scientists to configure how microscope samples are scanned and to control how the images from these scans are analysed by image processing and deep learning techniques.
Jordan presents Marvin, the first automated microscope product built on this platform and discussing the underlying technology. Marvin is paranoid about asbestos, he is able to perform analysis of air filter samples from asbestos air monitoring jobs 10x faster, and up to 50x higher volume than a human analyst of equivalent skill.
Taking AI to Production
By Dr Adam Makarucha
AI Technical Specialist, IBM
IBM Summit was recently officially announced as the world’s most powerful super computer. At 122 petaflops it provides unprecedented performance, which is largely due to its extensive use of GPUs and unique technology linking CPU and GPU. GPU technology has been driving revolutionary research in molecular modelling, fluid dynamics and space research, but most front of mind is in AI.
Almost every company is trying to use machine learning (ML) somewhere in their business, but how do you take the models developed with ML and put them into production using a repeatable and scalable process?
In this talk Adam discusses how using cloud native technologies can enable you to build a lifecycle around ML model development, deployment and ongoing monitoring.
A Neural Network’s Help In Understanding Our Nearest Planetary Neighbour
By Dr Eriita Jones
Planetary & Space Scientist
Imagine a phenomenon that does not occur anywhere on planet Earth: metres thick deposits of carbon dioxide ice become transparent, allowing sunlight to pass through them and warm the dark sand down beneath the ice. The base of the ice warms up just enough to sublimate, causing an explosive eruption of CO2 gas back up through the surface of the ice and spouting huge jets of dust and dirt into the air. The debris is carried a short distance by the wind before settling back down onto the surface. No human has even witnessed this springtime drama at the southern polar terrain of Mars, but satellites have captured many hundreds of high-resolution images of the dark fans and blotches left behind on the surface.
This talk will discuss how a neural network can find these features and help us understand them, how we can use machine learning to learn about the surfaces and subsurface of other planets, and even how neural networks can help us search for life elsewhere in the solar system.
The Future of Data Science
Senior Architect and Analytics Lead, LIFELENZ
One challenging aspect of machine learning - even for experienced practitioners - is choosing the correct tools for a specific task. Anyone using machine learning must make decisions about everything from how to clean data, what features to extract or build, or what learning algorithm to use. Recently, some groups have been looking to automate some parts of this process, or even the whole pipeline. This field is generally known as AutoML, and in this talk, Katherine gives a short introduction to an AutoML project out of MIT.
Big Data and Bioinformatics
Bioinformatician, Centre for Cancer Biology, University of South Australia
Single cell sequencing is a hot topic in biology at the moment. It allows the genes in thousands of individual cells to be analyzed separately. This can be very helpful if you want to distinguish cancerous cells from normal tissue, for example. Alongside this new technology comes massive amounts of data, leading to great technical and scientific challenges and opportunities for applying new machine learning techniques. Emily gives a brief intro to this topic - no prior biology knowledge necessary!
Athlete’s AI: Real-time video analytics in the hands of all athletes
By Dr. Mark McDonnell, Co-founder and CTO, Athlete's AI
Video analytics is a powerful tool for athletes and their coaches. It can help improve performance by providing biomechanical analysis, match statistics and tagging of key events. However, current solutions require painstaking manual labelling and editing, and so are high cost, slow and provide limited insights. Athlete’s AI combines computer vision, machine learning and cloud computing to deliver three distinct advantages over our competitors: lower cost, faster real-time analytics, and further insights.
In this presentation, Mark provides an overview of Athlete’s AI’s tennis product, and demonstrate its capabilities in action. He also describes some of the technical challenges that the team has solved by designing cutting-edge deep neural networks.