A room full of people got AI wrong, and it explained why credibility is now a commercial asset
In June I sat in the Bathurst library for two hours at an ABC workshop on media literacy, run with the Australian Library and Information Association and Western Sydney University.
Partway through, the facilitators ran an exercise. A series of images went up on the screen and the room was asked to say which were real photographs and which had been generated. Roughly two-thirds of the people present had already told the room they use AI tools in their work or study.
We scored about half. Which is to say, chance. A coin would have done as well.
I have thought about that number more than anything else from the session, because the people in that room were not naive about the technology. They use it. They still could not identify its output.
What the research says, in case the room was unusual
It was not.
A 2024 study led by Professor Sora Park and Associate Professor Tanya Notley, across the University of Canberra, Western Sydney University, QUT and RMIT, tested adult Australians on their ability to verify information online. Ninety-seven per cent showed poor or limited ability. Forty-five per cent showed no ability to identify misinformation at all. Three per cent demonstrated developed verification skills.
Ninety-three per cent said people need training in how to identify misinformation, which tells you something hopeful about public appetite and something bleak about current provision.
The workshop also walked through the #ArsonEmergency case from the 2019 and 2020 bushfires, researched at QUT by Tobias Keller, Timothy Graham, Daniel Angus and Axel Bruns. The claim that arson was behind most of the fires was false, and it was pushed deliberately and at scale while the fires were still burning. Not a rumour that got away from someone. A campaign.
The commercial reading, which nobody in the room was there for
I was the only person present, as far as I could tell, thinking about this as a business problem. Everybody else was there as a citizen, which is the point of the workshop and the right reason to attend.
But there is a consequence for anyone who needs to be believed in order to trade, and it is this.
When an audience cannot verify a claim themselves, they stop trying. That is not laziness, it is arithmetic: verification takes effort, the volume of material is now effectively infinite, and the effort does not scale. So people fall back on a shortcut, and the shortcut is to trust a source they already trust and let that source do the verifying.
Which means verification has become something you import rather than something you assert.
You cannot establish your own credibility by saying credible things, because saying credible things is now free and anybody can do it at volume. What you can do is have your claim carried by somebody the audience already trusts, who checked it first and whose own standing is on the line for having done so.
That is what earned media is. It has always been what earned media is. What has changed is that the alternatives have stopped working.
The part that should be uncomfortable
Everything a business produces about itself has become cheap to produce and therefore cheap to discount.
Your website copy, your case studies, your social posts, your carefully written explanation of what makes you different. All of it can now be generated in seconds, by anyone, at any quality level, in any quantity. Your audience knows this. They may not articulate it, but they are discounting accordingly.
Scarcity has moved. It used to sit in production, in the ability to make a polished thing. It now sits in verification, in the ability to demonstrate that a third party with something to lose checked and agreed.
This is why I am wary of the argument that businesses should simply publish more, faster, now that the tools make it easy. Publishing more of what an audience has learned to discount is not a strategy. It is a treadmill.
What the workshop actually recommended
For fact-checking, the facilitators taught the ESCAPE method: check the Evidence behind a claim, the Sources and who made it, the Context it originally appeared in, the Audience and Purpose it was built for, and the Presentation for signs of generation or logical gaps.
Senior ABC journalist Matt Martino made a point I thought was the most practical of the day. Do not attempt to verify everything, because you will exhaust yourself and give up entirely. Spend the effort on claims that carry consequences, and ignore the rest.
That is good advice for a reader. Reversed, it is also good advice for anyone trying to be believed. Your audience is rationing their scepticism. They will spend it on the claims that matter to them, which are precisely the claims your business most needs to survive scrutiny.
Where this leaves us
Australia has moved decisively on some things and not at all on others. Social media accounts for under-sixteens are banned. There is no equivalent mandatory safety regulation for AI systems, a gap the digital rights advocate Lizzie O'Shea addressed directly at the session, drawing the comparison with how vehicle safety standards eventually arrived through public pressure rather than industry initiative.
That gap will close at some point. In the meantime, the environment is what it is, and the businesses that will be believed in it are the ones that can point to somebody other than themselves who checked.
I went to that workshop to understand what my community is being told about trust. I came out with a sharper commercial argument than I went in with, which was not the plan.
If your audience cannot verify you, and increasingly they cannot, then the only claim that carries weight is the one you did not make yourself.
If you want to work out what your business can actually have verified, and by whom, that is the ground The Breakthrough Brief covers.
Sources
You can still catch the full Matter of Facts series on ABC iView.
Park, Sora, Tanya Notley, et al. Online Misinformation in Australia: Adults' Experiences, Abilities and Responses. University of Canberra, Western Sydney University, QUT and RMIT, 2024. Keller, Tobias R., Timothy Graham, Daniel Angus and Axel Bruns. #ArsonEmergency: Climate Change Disinformation During the Australian Bushfire Season 2019–2020. QUT Digital Observatory, 2020. Both verified 29 July 2026. The workshop-floor figures (the roughly 51% score on the image exercise, and around two thirds of the room already using AI) are Jess's own first-hand observation from the session and should be framed that way, not as published research.