Last week the Australian Federal Government released the 'Voluntary Artificial Intelligence (AI) Safety Standard', a guideline detailing 10 suggested AI guardrails and instructions for how to use them effectively.
“The guardrails help organisations to benefit from AI while mitigating and managing the risks that AI may pose to organisations, people and groups,” the government release states.
Artificial Intelligence experts from RMIT University are unconvinced the standard will have much of an impact, describing it as a ‘move in the right direction’ but ‘ambiguous’.
Balancing safeguards and red tape
“We know it’s important to implement safeguards as AI becomes more common – it's definitely a move in the right direction,” RMIT professor of business analytics and director of Enterprise AI and Data Analytics Hub research centre Kok-Leong Ong said.
"However, these new voluntary measures may not be suitably effective, while mandatory measures may just add another level of red tape.
“The voluntary approach means some companies may be selective about how they apply the safeguards and it’s ambiguous, leaving a business to assess risk.
“Another big issue relates to workforce, which is just not adequately trained to implement the proposed AI safeguards right now.
“There is also the balance between safety and speed for customers. For example, requiring disclosure of AI use may disrupt current effective credit card application processes.”
An ABC survey recently revealed a third of businesses using AI do not inform their employees or customers that they are doing so.
The research also revealed half of businesses using AI had not carried out human rights or risk assessments on their AI use and had no plans to do so.
Current standards insufficient
The Federal Government is working on a more robust set of rules for AI use in the Australian economy.
The government has set out three options for how those rules could be imposed, ranging from simply adjusting current digital software use laws, to creating a standalone, economy-wide act that could outright prohibit the use of AI tools deemed too dangerous.
One of the measures being considered could require 'developers and deployers' of AI to actively inform Australians when tools are used to make decisions about them, or when they are directly interacting with AI or AI-generated content.
“The voluntary standards are a welcome interim measure,” RMIT professor of information sciences and director of RMIT Centre for Human-AI Information Environments and the Social Change Lisa Given said.
“However, mandatory guardrails are needed to ensure appropriate protections for consumers, employees and others. They would also align Australia with other jurisdictions, such as the European Union.
“The voluntary standards will help organisations and regulatory bodies take the next step to ensure AI benefits the community.
“Especially given the rising challenges relating to the transparency in its design, application and use.
“The government is also looking to provide more clarity about what constitutes ‘high-risk’ AI, which is another critical issue that must be addressed.”
Addressing public concerns
In March the University of Queensland surveyed Australians on their attitudes toward AI and its uses in society.
Researchers found Australians a deeply divided on the topic of AI – 40% surveyed support the development of AI technologies, while 30% oppose it.
We also can’t agree on whether we think it will be a net positive overall – 40% say yes and 40% say no.
What we can agree on, is that we want the government to prioritise safety above all else.
A whopping 90% of those surveyed said they believe the government should create a new regulatory body to oversee AI specifically.
Federal Industry Minister Ed Husic said the government was working to address the public’s concerns and encouraged business to take active steps with the same goal.
"Australians know AI can do great things but people want to know there are protections in place if things go off the rails," he said.
"The public is concerned ... it is in the interest of business to act on those concerns.”