Today the Australian Department of Defence released ‘A Method for Ethical AI in Defence’ an outcome of a workshop in 2019 attended by over 100 representatives from Defence, other Australian government agencies, industry, academia, international organisations and media. The workshop was facilitated by Defence Science & Technology Group, RAAF Plan Jericho and Trusted Autonomous Systems Defence Cooperative Research Centre. Defence note that the report outlines a pragmatic ethical methodology for communication between software engineers, integrators and operators during the development and operation of Artificial Intelligence (AI) projects in Defence.

Trusted Autonomous Systems CEO Professor Jason Scholz said ”Trusted Autonomous Systems are very pleased to partner with Defence on this critical issue of ethics in AI. Ethics is a fundamental consideration across the game-changing Projects that TAS are bringing together with Defence, Industry and Research Institutions.”

AI and human machine teaming will be a key capability in the future of Australian Defence systems. Chief Defence Scientist Tanya Monro notes “…AI technologies offer many benefits such as saving lives by removing humans from high-threat environments and improving Australian advantage by providing more in-depth and faster situational awareness”.

Air Vice-Marshal Cath Roberts, Head of Air Force Capability said “artificial intelligence and human-machine teaming will play a pivotal role for air and space power into the future… We need to ensure that ethical, moral and legal issues are resolved at the same pace as the technology is developed. This paper is useful in suggesting consideration of ethical issues that may arise to ensure responsibility for AI systems within traceable systems of control”. These comments are equally important to the other service arms.

In 2019, the Trusted Autonomous Systems Defence CRC (TASDCRC) commenced a six-year Programme on the Ethics and Law of Trusted Autonomous Systems valued at $9M. Over the past two years the activity has conducted workshops, engagements and consultation with participants and stakeholders of the Centre, contributing to ADF strategy, producing diverse publications and influencing the design of trusted autonomous systems such as the game-changing Athena AI ethical and legal decision support system.

From 2021 the Ethics Uplift Program (EUP) of the TASDCRC will offer ongoing assistance to Centre participants through consultation, advice and policy development, supported by case analysis, education and enculturation

The Trusted Autonomous Systems affiliate researchers and employees participate in a wide range of events in consideration of the ethics and law of RAS-AI such as  ICRC, UNIDIR SIPRI, and NATO.

TASDCRC is a non-government participant in the United Nations (UN) Group of Governmental Experts (GGE) on Lethal Autonomous Systems (LAWS) to ensure the development of autonomous systems accord with ethical principles, the laws of armed conflict (LOAC) and in abidance with Article 36 weapons reviews.

The Defence Media Release reinforced that “The ethics of AI and autonomous systems is an ongoing priority and Defence is committed to developing, communicating, applying and evolving ethical AI frameworks”. Trusted Autonomous Systems are a partner to Defence on that journey. More details at https://www.dst.defence.gov.au/publication/method-ethical-ai-defence

Trusted Autonomous Systems CEO, Professor Jason Scholz was awarded the 2020 McNeil Prize today by Chief of Navy, Vice Admiral Michael Noonan, AO in a virtual ceremony.

In 2016, the Australian Naval Institute (ANI) created an award to honour an individual or individuals from Australian defence industry who have made an outstanding contribution to the capabilities and sustainment of the Royal Australian Navy (RAN). This award was named the McNeil Prize in honour of Rear Admiral Percival McNeil CB RAN (1883-1951).

The contributions of Prof. Scholz to RAN capability are articulated in the ANI Media Release and the ceremony underscored the importance of this field of research and contribution to future capability. Read the ANI Media Release here.

Congratulations Jason!

In identifying the key disruptive technologies of the future, drones appear on most major industry assessments as a top disrupter. Around the world, countries are scrambling to lead these developments with Queensland and Australia leading innovators and investors in drones, robotics and autonomous systems. Applications for drones exist in industries such as offshore energy, defence and science for “dull, dirty, dangerous and distant” operations. Independent industry research estimates the global drone market is expected to grow to $100 billion by the end of 2020.

But there is a problem.

The problem is that current assurance and certification processes are not well suited to unmanned systems and can take many months or even years. Coupled with low sales volumes, uncertainty and delays can significantly stifle growth and adoption. The accreditation of robotics and autonomous systems is done on a case by case basis and often by exception rather than streamlined.

To address this challenge, Biarri, Queensland University of Technology (QUT) and the Trusted Autonomous Systems Defence Cooperative Research Centre (TASDCRC), with financial support from Advance Queensland (the TASDCRC and this project receives funding support from the Queensland Government) and domain input from QinetiQ and Australian regulatory authorities, are designing software tools to accelerate and simplify the assurance and certification process for AI enabled and unmanned systems. These types of software tools have recently revolutionised many other industries and so the timing is right to accelerate innovation in the regulation industry via digital transformation.

By enabling faster and more efficient assurance and accreditation processes, our platform will facilitate innovation in a way which will catalyse the next wave of autonomous systems to solve large scale global challenges.

Why Now?

AI and autonomy are introducing new challenges for regulators, insurers, technology developers and system operators – with current methods, policies and practices lagging behind technical advances. Regulatory frameworks in Australia for domains like air and sea represent a mostly static, top-down regulatory compliance model of a public authority regulating under a legislative scheme.

As companies try to certify their new unmanned autonomous systems, facing uncertainty and lengthy delay, it is clear that we require new approaches to assuring land, maritime and air based pilotless vehicles, vessels and aircraft that are sustainable and effective. Owners and operators need to be able to update software and hardware to meet changing operational and regulatory requirements on the fly and not have to resubmit their systems for regulatory reapproval when they do.

The tools that will address this are part of the #RegTech digital transformation, which is based on changing the regulation processes via third generation computational regulatory regimes using digital technologies within digital environments. Such systems already exist in the finance industry, e.g. regulatory sandboxes established for fintech innovators (Allen, 2019; Piri, 2019).

A similar, flexible regulatory environment to what has been done in fintech and the supporting tools will enable the faster assurance and certification of aircrafts, vessels and vehicles driving greater innovation without compromising safety. Although potentially game-changing, building digital and physical sandboxes to work within traditional state-centred regulation is difficult and so it requires the specialised skills – skills which Biarri and QUT, in combination with its project partners, will bring to this endeavour.

What is the goal?

This project is seeking to answer the above challenges by bringing together world leading expertise from a number of partners such as Biarri and QUT’s Faculty of Law as well as experts in the appropriate regulatory environments via the TASDCRC. The goal is to create an integrated system of agile regulatory methods and AI audit tools, in order to simplify and streamline the interaction between the regulator and the regulated. These tools will dramatically shorten the current time it takes to assure and accredit unmanned systems, without compromising trust or safety, therefore providing a high level of value for businesses who need to quickly prototype unmanned systems that incorporate AI core capability.

This two-sided digital platform will also enable the operational parameters and software of unmanned autonomous systems to be rapidly adapted in response to changes in regulatory, legal, ethical and societal needs. The developed software will accelerate innovation and positive commercial outcomes by lowering the barrier to deployment for companies developing unmanned autonomous systems – without compromising trust and safety.

 

Anticipated Outcomes and Benefits

We expect there to be a number of benefits from the new digital platform. They include:
● World leading capability to deliver faster product development lifecycles for companies to safely test AI enabled unmanned autonomous systems.
● A world-first two-sided digital tool that integrates regulators with the regulated to increase the speed for organisations to reach a safe and trusted deployable state.
● A digital toolkit to allow the auditing and tracking of autonomous and AI based developments and simplification of reporting requirements for companies.

With these benefits Australia has the potential to lead the way in defining and benefitting from the billions of dollars of investments and commercial activities around autonomous systems.

As investment in AI robotics and autonomous systems ramps up, it is vital that these systems can be tested and assured to meet society’s expectations for their safety, trust and reliability.

However, there are no established standards systems or formal comprehensive codes of practice to accredit unpiloted, unmanned autonomous systems. Operations are generally permissible only under highly conservative constraints, authorised as an exception waiver by CASA, AMSA, DASA etc.

The lack of accreditation standards or codes of practice represents lost opportunity for innovators, for example, according to a US report (Jenkins & Vasigh, 2016) ‘every year that integration [of drones] is delayed, the United States loses more than $10 billion in potential economic impact. This translates to loss of $27.6 million per day that UAS are not integrated into the NAS [national air space]’.

The reason why there are so few standards, best practices or test and evaluation protocols for autonomous systems is a chicken-and-egg problem. The impetus to establish and maintain technical and social licence for autonomous systems is driven by end user demand. But end user demand is dependent on trusting technical standards and social licence to operate. Breaking out of this chicken-and-egg problem to advance the autonomous systems pipeline requires collaborative investment—beyond the reach of individual autonomous systems developers and business users.

The development of best practice policy, appropriate standards, and a strong accreditation culture has the potential to enhance innovation and market growth for drones with autonomous abilities.

Queensland is leading the way by investing in Australia’s capacity for translating autonomous systems innovation into deployments through the Trusted Autonomous Systems Defence CRC. To this end Directors of Autonomy Accreditation in Air, Land and Maritime domains will develop a national body of knowledge including methods, policies, and practices to support accreditation. Directors will address issues experienced by regulators, insurers, and technology developers by producing consistent (yet flexible) parameters for safe and trusted operations and improved agility to meet fast-changing technical and social licence needs.

It is with great pleasure that the Centre can announce the appointment of Rachel Horne as Director of Autonomy Accreditation – Maritime, effective Monday 17 August.

Rachel brings a wealth of experience and expertise in Australia’s maritime regulatory framework. She joins us from the Australian Maritime Safety Authority, where she has spent the last eight years in legal, regulatory and policy teams providing advice and managing projects aimed at improving the domestic regulatory framework.

Rachel is a subject matter expert on the domestic regulation of remotely operated and autonomous vessels and is invested in working to mature the assurance and accreditation frameworks available to better facilitate the safe and efficient update of autonomous technologies. The benefits this technology will bring from a safety, environmental, and efficiency perspective make this an important and timely undertaking.  More on Rachel on LinkedIn

The Centre is currently advertising for Directors in Air and Land domains, more information here.

Reference:

Darryl Jenkins & Bijan Vasigh, Commercial Drone Use Will Benefit the US Economy, from Drones Ed by Tamara Thompson, Page 150. Jan 2016. Greenhaven Press.