At a time of extraordinary transformation, it is vitally important that we ask the right questions about AI. What serves our economy, our democracy, our businesses, our students, our health system, our research base, our energy transition and our future sovereignty?
Professor Brungs gave this address at a UNSW Lunch at the Business Sydney Events Centre Wednesday, 3 June 2026
Can I begin by acknowledging the Traditional Custodians of the land on which we meet, the Gadigal people of the Eora Nation. I pay my respects to Elders past and present, and acknowledge all Aboriginal and Torres Strait Islander people with us today.
It is a great pleasure to be here with Business Sydney, and I want to thank Paul Nicolaou and the Business Sydney team for so generously hosting us.
UNSW has invited this group because much has been discussed about the urgency of artificial intelligence as an issue for business, universities, government and the community. But when all is said and done, a lot more has been said than done. At UNSW we believe it’s a time for doing, and I’m guessing you are here because you agree.
What a time to be alive.
Last week Sam Altman, of OpenAI, beamed in to a CBA conference to say that he may have been wrong about some of the social and economic implications of AI. According to Sam, we are not heading towards a “global jobs apocalypse”.
Only in the Age of AI could an industry leader announce, with a straight face: my technology will not see the biblical four horsemen — conquest, war, famine and death — ride through the white-collar workforce and bring on the end of the world. What a relief.
And by the way, build more data centres.
Naturally, Sam Altman was contradicted by Dario Amodei from Anthropic, who doubled down on the jobs apocalypse: AI could eliminate a very large share of entry-level white-collar jobs. And by the way, build more data centres.
I’ll leave the tech billionaires to disagree and push their own self-interest. There is no doubt we are not dealing with a small operational upgrade. We are dealing with a technology that is forcing business leaders, educators, policymakers and communities to ask very large questions, very quickly.
Asking the right questions is as important as our actions.
One of the defining and occasionally frustrating characteristics of academics is that we devote significant effort to properly framing the question before we will even contemplate looking for the answer.
In the case of AI this is the right approach. To get the best outcomes we need to ask the right questions.
So the question I want to focus on today: how do we make sure this technology expands human capability rather than narrowing opportunity?
At UNSW, our focus is to respond to the AI transformation with expertise, responsibility and purpose.
Our University’s strategy, Progress for All, is built around a simple but demanding proposition: progress is only meaningful if it is shared. UNSW’s mission is to advance Progress for All globally through transformative education, innovative research and meaningful engagement with people, communities and partners in Australia and around the world.
In the age of AI, that purpose becomes even more critical. AI is a force multiplier.
Left unchecked, it will also multiply inequality. That is my number one concern about AI, and it should be yours.
The most important AI question is not whether my iPhone can summarise my emails. It is who gets locked out of opportunity and why.
I intend to back up my thesis with two main sources: Pope Leo XIV and the Wharton School of Economics. It’s up to you which one you consider authoritative.
Pope Leo XIV’s encyclical, Magnifica Humanitas, is a powerful declaration of the risks inherent in AI. It is not anti-technology. It does not say we should abandon the AI project.
But it does say something profound about power.
It warns that “those who control AI will impose their own moral vision”. It says that merely regulating AI is not enough; “it must be disarmed”.
That is very strong language. It calls on us to pause and ask the right questions. What are the moral implications of AI? What are the implications for our humanity?
It reminds us that AI is not just a tool. It is becoming an environment. A layer of decision-making, mediation and interpretation sitting between citizens and institutions, workers and employers, patients and health systems, students and knowledge.
The people who build that layer will shape the world that sits inside it.
And here is where Australia needs to be very clear-eyed.
The Silicon Valley billionaires urging Australia to move fast, build data centres, lower barriers and become an AI powerhouse are not neutral observers of our national interest. They have an agenda.
That does not make them wrong. It absolutely does not mean we should reject investment, technology or ambition. Australia should absolutely build sovereign AI capability. We should be serious players, not passive consumers. We should not become, as Andrew Charlton has put it, permanent renters of intelligence.
But nor should we confuse the commercial interests of global technology firms with the long-term interests of Australia.
They may want Australia as a stable, energy-rich, English-speaking platform for AI expansion. Fine. Our responsibility is to ask: what serves Australia and how do we make sure it does?
The right question is: what serves Australia?
What serves our people, our economy, our businesses, our students, our workers, our democracy, our health system, our research base, our energy transition and our future sovereignty?
That is not a question Silicon Valley can answer for us. Nor would they even ask.
It is a question for Australians in leadership: in business, government, education, finance, philanthropy and civil society.
Most of you are not dealing with AI in the abstract. You are dealing with it day by day: in your operations, your customer models, your risk systems, your workforce planning, your cyber settings, your legal exposure, your budgets and your board conversations.
You are being asked to move quickly and carefully at the same time.
Move too slowly, and you fall behind. Move too quickly, without judgement, and you risk harm, mistrust and wasted investment.
So the challenge is intelligent adoption. And intelligent adoption requires the right questions and deep human expertise.
One thing we have all learned about generative AI is that it produces outputs that are credible and plausible even when they are 100 per cent wrong.
AI is like the world’s most confident intern: fast, energetic, enthusiastic — and wrong with complete conviction.
We joke about it until the wrong output is a legal opinion, a financial model, a medical summary, a cybersecurity assessment, or a piece of code inside critical infrastructure.
The danger is not that AI gives obviously poor answers. The danger is that it gives answers that are fluent, structured and convincing — but false.
It takes human expertise to recognise that. It takes an experienced lawyer to see why a plausible argument fails. It takes an experienced accountant to see why a model is elegant but wrong. It takes an experienced software engineer to see why code works in one case but fails dangerously in another. It takes an experienced analyst to know which assumptions are doing the real work.
And that brings us to one of the central paradoxes of AI and work.
Many of the tasks most exposed to AI are the very tasks through which graduates have traditionally learned.
The functional work. The first drafts. The basic analysis. The document review. The coding tasks.
The work that can seem routine from the top of an organisation, but which is formative at the beginning of a career.
If that work is replaced wholesale by AI agents, where do the future experts come from?
Who develops the judgement to supervise the machine? Who watches the watchers?
I’m not being rhetorical. It is one of the labour-market questions of the next decade.
At UNSW, we live this challenge every day. Our students will graduate into a world where AI is normal. So the answer cannot be to pretend it does not exist. Nor can it be to treat AI only as a threat to academic integrity.
Integrity matters deeply. A UNSW qualification stands for capability, effort, knowledge, judgement and trust.
But we cannot simply police misuse. Our duty is to prepare students to use AI well: critically, ethically, creatively and responsibly.
That is why UNSW has been developing what we describe as a Can, Can’t, Must approach to assessment in the age of AI.
There are things students can use AI for, because they reflect the world they are entering.
There are times students can’t use AI, because it would undermine the purpose of assessment and the integrity of the qualification.
And there are parts of the curriculum where students must use AI, so they graduate capable of using technology to expand their own capability and operate in the world as it is.
Every organisation will probably need its own version of Can, Can’t and Must. This is where the skills conversation becomes central.
El Iza Mohamedou, Head of the OECD Centre for Skills, recently spoke to Universities Australia about the changing relationship between education, skills and work. Her message was clear: the labour market is moving away from a narrow reliance on formal qualifications and towards a sharper focus on demonstrable skills and capabilities.
That is exactly what we are seeing. The more powerful AI becomes, the more important human capabilities become.
That is why UNSW is embedding future-focused and AI-relevant skills into our education offerings, refreshing graduate attributes, and exploring a skills passport so students can evidence their capabilities to employers.
AI, used properly, can help us solve tough problems.
Harrison.ai, an Australian health technology company co-founded by UNSW alumni Dimitry and Aengus Tran. Its AI-powered diagnostic support tools are now being adopted internationally across radiology and pathology.
The Tran brothers thought carefully about the role of AI, the role of humans and the needs of their customers.
That is the kind of Australian AI story we should want more of.
Not just importing someone else’s model or renting intelligence, but building capability here, with Australian talent, connected to Australian institutions, solving real problems and exporting value to the world.
Dr Sue Keay, head of our AI Institute, sees a future where domain-specific, smaller AI models replace ChatGPT and Claude. Providing an opportunity for Australia to compete or even lead in the field.
The Wharton paper, The AI Layoff Trap, makes the economic warning clear. In a competitive model, each firm may act rationally by automating to reduce costs. But if many firms do that at scale, they may collectively erode the consumer demand on which they all depend.
It suggests that what is rational for one firm may be destructive if everyone does it at once.
It is the economic equivalent of everyone standing up at a concert to get a better view. One person improves their position. When everyone does it, no one sees better, and everyone is less comfortable.
I don’t suggest that businesses should ignore productivity. Productivity matters enormously. Australia needs productivity growth.
My last CBD lunch was to release a UNSW occasional paper, An Australian Productivity Roadmap for Policymakers, that set out in detail the terrible cost to young Australians if we fail to fix our anaemic productivity growth.
AI will create real efficiencies, real new markets, real improvements in service and real opportunities for growth.
But asking the wrong questions will lead to a narrow conception of efficiency that can become self-defeating.
If AI adoption hollows out entry-level career paths, weakens consumer demand, concentrates income and reduces the shared base of capability in the economy, then the long-term effects may be very different from the short-term gains.
The clearest pathway to avoiding the worst harms of AI is deeper engagement between universities, industry, finance, corporate Australia, government and community.
Industry brings the live problems.
You know where productivity is constrained. You know where customers are frustrated. You know where systems are slow, expensive or fragile. You know where AI is already having an impact and where it is overpromising.
Universities bring a different kind of contribution.
We bring research depth, global networks, and the ability to convene disciplines that do not always sit together: computer science, business, law, psychology, design, education, engineering, medicine, ethics, economics and public policy.
Professor Toby Walsh is one of Australia’s leading voices on artificial intelligence. He has been reminding us for many years that AI is not magic. It is technology built by people, trained on human data, deployed in human institutions and governed by human questions and choices.
UNSW Business School has a critical role in this conversation.
I want to acknowledge Professor Paul Andon, newly appointed Dean of UNSW Business School, who will host the panel discussion after Tarushi Nandwani’s remarks.
UNSW Business School is a global leader. It is also actively rising to the challenge of the AI future.
And you will hear shortly from Tarushi, a UNSW student in Data Science and Decisions and former President of the UNSW AI Society.
I am especially pleased Tarushi is speaking today because students often understand the practical reality of this transition before institutions do.
For students, AI is already in the study process, the job market, the workplace, the social world and the information environment.
No UNSW student believes AI means we no longer need to learn.
Because the more powerful the tools, the more important it is to know what questions to ask, what evidence to trust, what assumptions to test and when a human being needs to intervene.
The next generation will not thrive by outsourcing thinking.
They will thrive by using new tools to think better.
My invitation today is direct. Partner with us.
Work with our academics. Work with UNSW Business School. Work with UNSW Employability. Work with our students and graduates.
Bring us the problems that matter. Challenge us to be useful.
Help us ask the right questions by understanding the capabilities your organisations need, not in the abstract, but in practice.
And work with us to make sure we do not accidentally remove the career pathways through which future experts are formed.
Because AI will shape the future. But it will not decide the future on its own. We will.
And the choices we make now — as universities, businesses, governments, civic institutions and leaders — will determine whether AI becomes a tool that concentrates advantage, or a platform for broader capability, shared prosperity and human progress.
