It has become common for employees to use AI to assist with work-related communication such as emails. (ABC News: Peter Mullins)
The first email was sent in 1971 as an experiment to see if two computers could exchange messages.
Since then, the communication tool has become commonplace in businesses, but with the rise of artificial intelligence (AI), those exchanges have become even less human.
Griffith University's Mai Nguyen said it was now the norm for workers to use AI to sound more professional and use the right grammar.
But the business school senior lecturer warned that using the wrong prompt in an AI generator could exaggerate facts and make the text's tone inappropriate.
'Couched in accusatory language'
The widespread use of AI had recently prompted Australia's workplace tribunal to order an employee and their workplace to cease using the technology in all future correspondence.
The Fair Work Commission recently ordered an employee and their workplace to stop communicating with AI.(ABC News)
The order was handed down in a bullying complaint before the Fair Work Commission in late July.
An employee working as a caretaker for a Queensland body corporate submitted to the tribunal that the workplace repeatedly refused to clarify their duties, provided confusing and shifting directions, and made several other allegations.
Commissioner Sarah McKinnon ruled that the employee had been bullied.
But she noted emails from the employee and the workplace and found the caretaker's correspondence was unhelpful and appeared to have been prepared with the use of an AI tool.
AI can be used to check grammar and spelling and to increase efficiency in the workplace. (ABC News: Greg Bigelow)
Commissioner McKinnon described the communication as lengthy, wide-ranging, replete with generalisations, repetitive, and often couched in accusatory language.
She said it was unsurprising that the workplace eventually chose not to engage with the employee.
Unfortunately, its choice to 'disengage and dig in', my description, had the result that it did not engage in any meaningful way with many of the concerns [the employee] was raising, Commissioner McKinnon said.
This was to the detriment of all parties.
To prevent further bullying of the employee, Commissioner McKinnon ordered all future communication between the parties be accurate and respectful.
There is to be no use of artificial intelligence tools in the preparation of correspondence between the parties, she said.
Parties should endeavour to keep their communications brief and to the point.
The order was to take effect immediately.
Digital communication Professor Daniel Angus said it was an interesting case.
The Queensland University of Technology professor said there was a natural appeal to want to use the technology to make text more forceful or persuasive.
He said someone could use it to overwhelm an opponent within a debate in this kind of adversarial environment.
Should you disclose an email has been written with AI?
Professor Angus said we need to normalise disclosing the use of AI a lot more but conceded it was a really tricky topic.
What we're seeing is that there are some folks who are vehemently opposed to the use of AI, he said.
You can understand why people don't want to be forthcoming about their use of AI.
Law research fellow Guzyal Hill, from the University of Melbourne, said very few organisations had AI policies but said there was a growing push for it.
It's difficult to avoid AI in terms of large language models in the modern society, she said.
Australia does not have an AI Act, so each organisation has to develop its own AI Strategy/AI Policy.
The AI Governance Chair at HWLE Lawyers said there wasn't much harm in using AI for emails as long as it was used skilfully, in line with legislation and internal work policies.
But she said it should be discouraged if private and confidential information was being put into the AI generator.
The caution is always about the way it's used, Dr Hill said.
Can you tell if AI wrote an email?
Professor Angus said there were telltale signs when AI has written texts, but that humans were notoriously bad at picking it up.
I often think about the maligned em-dash (—), which is a real victim of the AI moment, he said.
Professor Angus said the tone of voice was another revealing sign of the AI.
You could read something and think, 'Well this doesn't sound like them,' he said.
Professor Angus warned about the increasing outsourcing of emails and other forms of communication to AI agents.
He said that would put workplaces in a really weird and almost dystopian position where it is just a bunch of AI agents talking to each other while the workers are seemingly unaware of what they're talking about.
Dr Nguyen reads up to 30 emails each day, and she suspects about 80 per cent of them are AI-generated.
Especially the long emails, I can predict that [they] come from some sort of AI, she said.
If using the technology for emails, Dr Nguyen advised double-checking that the text is clear, precise, and what you intended to send.
At the end of the day, the email comes from the name of the sender, she said.
