The Australian Government aims to elevate its digital economy and society to the top 10 by 2030 through its Digital Economy Strategy. To achieve this vision, the country must implement modern and intelligent systems and AI regulations that keep it at the forefront of emerging technologies.
Artificial Intelligence (AI) and Automated Decision Making (ADM)
Artificial Intelligence (AI) and Automated Decision-Making (ADM) offer significant potential for productivity, economic growth, and improved living standards. However, outdated legislation and a poor understanding of requirements can hinder their adoption, ultimately undermining public trust and confidence. Therefore, it is crucial to ensure that regulatory systems remain fit for purpose and agile to maximize opportunities and manage risks effectively.
AI and ADM are already being used across the economy and society, solving problems like brain aneurysms and fruit orchard management. Governments are also utilizing these technologies to improve service standards and operational capabilities, implementing smart gates at international airports and using AI to detect mobile phone usage by drivers.
With increasing global investment and product development, Australia must embrace these technologies to maintain productivity and living standards. To do this, the country must mobilize its collective capabilities, talent, and resources to be early adopters and significant contributors to the progress and regulation of these technologies.
By modernising regulatory settings and systems, Australia can ensure they are fit for the digital age and promote the responsible uptake of new technologies.
Positioning Australia as a leader in digital economy regulation will enhance public trust and confidence, facilitating greater uptake of these technologies in the long term.
Modernised legal frameworks will provide certainty about the benefits and risks of adopting these technologies, encouraging increased uptake and investment.
Additionally, consumers will have confidence that AI and ADM are being deployed in a trusted, secure, and beneficial manner.
AI Ethics Principles
To strike a balance between innovation and safeguarding society. Australia has 8 Artificial Intelligence (AI) Ethics Principles, designed to ensure AI is secure, safe, and reliable.
They will help:
- achieve safer, more reliable and fairer outcomes for all Australians.
- reduce the risk of negative impact on those affected by AI applications.
- businesses and governments to practice the highest ethical standards when designing, developing and implementing AI.
1) Human, Social and Environmental Wellbeing
This principle emphasises that AI systems should be employed for positive results that benefit individuals, society, and the environment. The goals of AI systems must be clearly defined and justified. AI systems that address global issues such as the United Nations’ Sustainable Development Goals should be promoted.
Ideally, AI should be used to benefit all humans, including future generations.
AI systems designed for legitimate internal business purposes, such as improving efficiency, can have significant impacts on individual, social, and environmental well-being.
These impacts, both positive and negative, must be considered throughout the lifecycle of the AI system, including any impact outside the organisation.
2) Human-centred Values
This principle aims to ensure that AI systems are in line with human values. The machines should work for humans and not the other way around. These systems should promote an equitable and democratic society by respecting, protecting, and promoting human rights, enabling diversity, respecting human freedom and individual autonomy, and safeguarding the environment.
It is important to carefully consider the risks to human rights, as AI systems can both enable and hinder fundamental rights. It is acceptable to interfere with certain human rights, but only when it is reasonable, necessary, and proportionate.
All individuals interacting with AI systems should have complete and effective control over themselves. AI systems should not undermine the democratic process or take actions that threaten individual autonomy, such as deception, unfair manipulation, unjustified surveillance, or failing to maintain alignment between a disclosed purpose and actual action.
AI systems should be designed to enhance, complement, and empower human cognitive, social, and cultural skills. It is recommended that organizations designing, developing, deploying, or operating AI systems hire staff from diverse backgrounds, cultures, and disciplines to ensure a wide range of perspectives and minimize the risk of overlooking important considerations that may only be noticeable to certain stakeholders.
3) Fairness
The principle of fairness in AI aims to ensure that AI systems are designed with the goal of inclusivity throughout their entire lifecycle. This means that AI systems should be focused on the user and designed in such a way that all people interacting with them can access related products or services. It is important to consult with stakeholders who may be affected by the AI system throughout its lifecycle and ensure that people receive equitable access and treatment.
This is particularly crucial because there are concerns about the potential for AI to perpetuate societal injustices and have a negative impact on vulnerable and underrepresented groups, such as those related to age, disability, race, sex, intersex status, gender identity, and sexual orientation. Therefore, measures should be taken to ensure that any decisions made by AI systems are compliant with anti-discrimination laws.
4) Privacy, Security and Protection
Our aim is to prioritise respect for privacy and data protection in the use of AI systems. To achieve this, we implement proper data governance and management throughout the entire lifespan of the AI system. For instance, we ensure privacy by anonymising data that is utilised by AI systems. Furthermore, we continuously assess the soundness of the connection between data and inferences drawn by AI systems.
Additionally, we take measures to ensure the appropriate security of both data and AI systems. This includes identifying potential security vulnerabilities and ensuring resilience to adversarial attacks. We account for unintended applications of AI systems and potential abuse risks and take the necessary steps to mitigate them.
5) Safety and Reliability
The aim of this principle is to guarantee that AI systems perform consistently in line with their intended purpose throughout their lifespan. This involves ensuring that AI systems are dependable, precise, and replicable as required. Additionally, AI systems should not present an undue safety hazard and should put in place safety measures that are commensurate with the scale of potential risks.
Regular monitoring and testing of AI systems should be conducted to ensure they continue to meet their intended purpose, and any issues identified should be addressed with continuous risk management as appropriate. Responsibility for ensuring that an AI system is robust and secure should be clearly and appropriately assigned.
6) Explainability and Transparency
Each stakeholder group considers it crucial to achieve transparency in AI systems through responsible disclosure for the following reasons:
- For users, whatever the system is undertaking and why.
- The systems’ operations and input data are made available to producers, such as those engaged in the assessment and certification of AI.
- Understanding operations and entering data is necessary for people installing and running the system.
- For an accident investigator, in case an accident occurs.
- In the context of investigations for regulators.
- For individuals involved in the judicial system, to guide evidence and judgement.
- To increase the public’s trust in the technology.
Responsible disclosures must be made on time and include plausible explanations for the results of AI systems. This comprises details that aid in understanding outcomes, such as crucial variables in decision-making.
This concept also tries to guarantee that individuals may request a meaningful disclosure about an AI system and learn when one is interacting with them (regardless of the impact).
7) Contestability
The purpose of this principle is to guarantee that individuals have access to effective and accessible mechanisms to challenge the use or output of an AI system. This is particularly important when the AI system significantly impacts a person, community, group or environment. The threshold for determining “significant impact” will vary depending on the context, impact, and application of the AI system.
It is crucial to establish public trust in AI by ensuring that there is a means of redress for harm when things go wrong. Special attention should be given to protecting vulnerable individuals or groups.
To make the process of contestability effective, there should be sufficient access to the information available to the algorithm and the inferences drawn from it. In situations where decisions significantly affect rights, there should be an effective system of oversight that makes appropriate use of human judgment.
8) Accountability
The responsibility of designing, developing, deploying, and operating AI systems lies with the relevant organizations and individuals. However, legal principles for holding them accountable are still in the development phase.
To ensure accountability and responsibility for AI systems and their outcomes, measures should be implemented before and after their design, development, deployment, and operation. The decision-makers must be identifiable, and the appropriate level of human control or oversight should be considered for a particular AI system or use case.
AI systems that significantly affect an individual’s rights should be accountable for external review. Independent oversight bodies should be provided with timely, accurate, and complete information for this purpose.
Our experienced information security consultant can help you understand and implement the AI regulations and AI Ethics with clarity to guarantee the security of data and systems.
Recent Comments