AI State of Mind

TLDR (too long; didn’t read):

• AI is applications or systems that use techniques to perform specific tasks and “automate” these tasks, replicating human-like intelligence. It’s math!

• Whilst AI has revolutionised industries, resulting in efficiencies, streamlined workflows and greater productivity, the lack of regulation in some countries raises concerns about its use.  

• The EU passed the world’s first legislation on regulating AI in August 2024, with Australia yet to follow suit.

• Risks of AI use include AI system hallucinations, unintended bias and overarching privacy and data concerns.

• In the absence of Australian legislation on AI use, businesses should ensure they are keeping up with industry best practice to safeguard their organisation and employees against risks.

What is AI?

Whilst “Artificial Intelligence” (‘AI’) does not have a globally uniform definition, it can loosely be defined as, applications or systems that use techniques to perform specific tasks and “automate” these tasks.

At the very basic definition, it’s math not magic.  

AI attempts to replicate human-like intelligence to perform simple, repetitive, or even complex and cognitive tasks.

The most popular and notorious form of AI is “generative AI” (‘GenAI’), which materializes through popular applications such as ChatGPT or CoPilot.  

Whilst this technology has proven to be revolutionary in industries, the concept of how to regulate such a piece of technology is rife globally.

AI Regulation in Australia & Globally

The European Union (‘EU’) has led the initiative in regulating AI by introducing the world’s first comprehensive law on AI, through the AI Act ('The Act’) in August of 2024.

This Act applies to all businesses operating in the EU, and all businesses globally that may offer their products or services to customers in the EU. This framework establishes rules and obligations for providers of AI systems and users, based on the perceived level of “risk” posed by the AI system.

Whilst Australia does not have specific legislation on the regulation of AI, in June 2024, the Government released the ‘National Framework for the Assurance of Artificial Intelligence in Government’.

This Framework is likely to become the foundation upon which AI regulation for Australia is based.

Risks of Use

The dangers and perceived risks when using AI include:

• Privacy and data concerns: as AI systems often collect large amounts of data, including personal information, there are concerns raised about the overall security of the data it holds and the lack of regulation in some countries.

• Hallucinations: it has been shown that an AI system can generate false and misleading outputs and information, where the user prompt or input is not understandable to the AI system due to insufficient or low-quality training data.

• Unintended bias: AI systems can tend to produce “biased results” that reflect human values in a society, rather than providing neutral and not-skewed outputs, which is crucial for a system of its kind.

Best Practice Strategies

In the absence of uniform law governing and regulating the use of AI in Australia, businesses should ensure they are adopting best practice strategies, through:

• Ensuring they have implemented an AI Policy within their organisation, aimed at equipping employees with guidelines on appropriate use and relevant risks, along with mitigation strategies.

• Instilling within employees, a “Trust but Verify Outputs” mindset – to ensure that any output generated by an AI tool is verified against a credible source, checking the information is accurate, appropriate and not biased.

• Mandate an “approved AI tool” organisation-wide and ensure due diligence has been conducted into the AI system to make an informed decision.

• Review of all contracts and service terms to include disclaimers for the use of AI tools being included in such service delivery.  

Adding to the complexity is the consideration of AI and implications under insurance products. Whilst insurers are welcoming the use of AI to minimise claims processing costs internally, and predict certain events or likelihood of events, they are yet to publish a stance on claims involving an AI tool mishap. It is likely this would fall into third party liability, however, carve outs and liability limitations may mean that AI could be excluded from any claim from a business if it is the cause of an otherwise insured event. A timely reminder for businesses to speak with their broker or insurer to ensure that the use of AI will not exclude them from the ability to claim.  

Businesses should ensure they are keeping in the know with Australian AI regulations as its prevalence and proliferation in society is only growing, along with the need for uniform legislation in Australia.

Disclaimer: This article is intended for informational purposes only and should not be interpreted as legal or financial advice. Lextech strongly recommends seeking professional advice to ensure that any actions taken are tailored to your specific business needs and circumstances. Lextech shall not be held liable for the accuracy or completeness of the information provided herein, particularly when accessed at a later date, as the content reflects the state of knowledge at the time of publication.

‍