As Australian businesses increasingly explore generative AI’s potential – aiming for boosted productivity and improved employee experiences – a significant portion of the workforce may be unknowingly embracing these tools while overlooking inherent risks.
Recent surveys underscore this trend. A January survey by Google, conducted in partnership with IPSOS, revealed that nearly half of all Australians are using generative AI, with almost 75% utilizing it for work-related purposes. Furthermore, a study from the HR platform Workday indicated that approximately 65% of Australian workers report their employer has already integrated AI into their workplace.
Even seemingly simple applications of generative AI, like composing business emails, can lead to unexpected and costly consequences. Michael, director of End Of Lease Cleaning Melbourne, experienced this firsthand, losing $2000 due to an error in a generated email.
“We were trying to save time by automating the creation of lists detailing our cleaning services,” he explained. Instead of manually typing out each service and its associated cost, his team began using a generative AI tool to produce these emails and quotes.
While this dramatically reduced their response time from five hours to just one, Michael admitted the system was far from flawless. The AI frequently misidentified cleaning requirements; for example, it often listed a ‘full wall clean’ instead of the requested ‘spot wall clean,’ failing to adjust the accompanying quote accordingly.
These errors resulted in his team being obligated to provide the more extensive service at the lower price point, leading to financial losses. The tipping point arrived in March when an AI-generated quote for a particularly “filthy” property (estimated at $2000) was sent to a customer containing numerous inaccuracies.
“We lost quite a lot of money,”
Michael stated, noting that the client ultimately went with another company due to the errors.
Michael is not alone; other Australian workers have reported similar issues, including AI tools mistakenly addressing customers, clients, and colleagues using incorrect names or titles—potentially damaging business relationships.
Robyn Chatwood and Michael Park, Intellectual Property and Information Technology lawyers at Dentons, caution that the risks extend beyond mere embarrassment. Generative AI tools are prone to “hallucinations,” fabricating information that isn’t factual. Incorporating such falsehoods into professional communications can create serious problems.
Furthermore, these tools can inadvertently infringe on copyright or violate confidentiality protocols, leading to potentially severe repercussions. As Chatwood emphasized, “You can’t just say the machine made a mistake; you have the responsibility and liability.”
Park advises Australian workers to adhere strictly to their employer’s AI usage policy, regardless of the temptation to expedite tasks using these tools.
“If your policy says don’t do it, then just don’t do it,” he stated. “You’re protecting yourself from potentially getting into trouble.”
Workers or small business owners lacking a formal AI usage policy should exercise extreme caution.
Following the $2000 loss, Michael and his team have discontinued using generative AI for all business correspondence. While their response time has returned to the original five-hour mark, he believes it’s a worthwhile trade-off compared to incurring further financial setbacks.
“If you are using AI, you definitely need to read everything two to three times before you send that email,” Michael concluded.
Key takeaways regarding generative AI use in the workplace:
- Be aware of the potential for errors and inaccuracies.
- Always review and verify AI-generated content before sending it.
- Adhere to your employer’s AI usage policy.
- If no policy exists, exercise extreme caution.
