Australian Formula One legend, Sir Jack Brabham, once observed that ‘good drivers can drive anything and can adapt to anything and that has never changed’.
His observation offers a lesson for Australia’s approach to Artificial Intelligence (AI).
We are accelerating towards an uncertain destination, yet Canberra appears increasingly concerned with controlling the steering wheel rather than understanding the road ahead.
History offers few encouraging examples of governments acquiring extensive power over information and then choosing not to exercise it.
The deeper problem was identified long before Large Language Models (LLMs) existed.
In 1941, Jorge Luis Borges published his short story, The Library of Babel, imagining a universe consisting of an apparently infinite library containing every possible combination of a fixed set of letters, spaces, and punctuation. Within its endless galleries existed true histories, coherent philosophies, and scientific discoveries – alongside every conceivable falsehood, contradiction, and meaningless string of characters. The library contained all possible answers, yet its inhabitants had no reliable means of identifying the correct ones. Their predicament was not scarcity of information but the impossibility of reliably distinguishing signal from noise.
LLM’s are not literal libraries of every possible book. They generate responses from patterns learned during training, and their fluency can mask uncertainty.
The intellectual parallel, however, is clear.
Borges imagined unlimited information without a reliable navigator. AI offers something that resembles one. The librarian has become the algorithm. This creates a potential democratisation of knowledge.
Citizens can interrogate legislation, test ministerial claims against published data, and examine the assumptions behind government policy. The LLM will not always be correct, but neither will the minister.
For the first time, the public’s capacity to scrutinise political authority may develop faster than the institutions that exercise it. The critical political question is therefore not whether the technology works, but who controls the navigator?
Australian cricket legend, Sir Donald Bradman, warned, ‘It is unwise for a batsman specifically to make up his mind before the ball is bowled to where he will hit it.’ Bradman’s observation offers another lesson for AI governance. Just as a batsman must judge the delivery before committing to a stroke, governments must understand technological change before determining how it should be regulated.
The question is whether Canberra is reading the delivery or swinging at a ball that has yet to be bowled.
In announcing the proposed AI standards, Prime Minister Anthony Albanese said they would ‘deliver greater clarity, speed up approvals and streamline the process for verifying compliance with energy, water, safety, and other requirements’.
The final two words merit attention – ‘other requirements’.
In poker, a tell is a small signal that can reveal more than intended. The Prime Minister’s phrasing is not evidence of a concealed agenda. It is, however, a political tell worth examining.
Energy and water constraints are legitimate, as data centres are resource-intensive. Safety is a standard responsibility of government. Consultation material already lists planning, community impacts, national security, and copyright. The more consequential question is whether future governments might extend regulation from physical infrastructure to intellectual output.
Could ‘other requirements’ include official definitions of misinformation, acceptable political discourse, or conclusions judged harmful?
These remain hypothetical, not descriptions of current Australian law. They nevertheless highlight a fundamental distinction. One domain concerns electricity, water, and buildings. The other concerns citizens, knowledge, and the distribution of political authority.
Consider one of the most elastic terms in contemporary political language – misinformation. Does it mean a demonstrably false statement? A deliberate lie? An honestly held but mistaken belief? Or an uncomfortable truth that contradicts an officially preferred narrative? A lie requires an intent to deceive. An error may not. An inconvenient truth may be entirely accurate. Combining these categories mistakes political acceptability for truthfulness. And who decides whether the government’s understanding of truth is correct?
Australian political life supplies a familiar ritual. A journalist asks an awkward question about expenditure or policy failure. The reply arrives with weary predictability – ‘I reject the premise of that question’. Occasionally the premise is genuinely false. More often the phrase functions as a rhetorical escape. The minister need not identify the faulty premise or offer evidence. The question is simply dismissed. An LLM offers a different possibility. A citizen can ask it to identify the premise, test the evidence, and assess whether the question was unreasonable. Unlike a press conference, the interrogation need not end when the minister decides enough time has been spent answering questions. The citizen can keep asking and asking and asking.
This is where Borges’s library becomes politically significant. Imagine a machine capable of locating the most relevant volumes in that infinite collection. Now imagine its administrators deciding which galleries the public may explore. Certain books are judged dangerous. Others are labelled misinformation or said to undermine confidence in our institutions. The administrators need not destroy anything. They need only prevent the navigator from directing readers to particular shelves. The books remain, but for most people they effectively disappear.
The greater risk may not be that an LLM refuses to answer, instead, it may become highly skilled at presenting carefully selected answers as the complete picture. The librarian reports that nothing relevant is on the shelf. Private technology companies are not neutral custodians. Their commercial incentives and institutional preferences deserve scrutiny.
Transferring intellectual authority from Silicon Valley to Canberra does not automatically solve the problem. It merely changes the librarian’s employer.
A liberal approach would be to preserve competition between systems, require transparency about restrictions, and distinguish the prevention of demonstrable harm from the suppression of lawful inquiry. Liberty depends on restraining concentrated power, not on trusting those who hold it to remain perpetually benevolent.
The deepest political consequence of AI may not be machines surpassing human intelligence. It may be that ordinary citizens become more intellectually independent than governments have traditionally expected or preferred. That is the delivery Canberra must learn to read.
Bradman understood that anticipating every delivery was impossible. Brabham understood that controlling a machine was no substitute for knowing how to drive it. Neither confused preparation with absolute control.
Before Canberra takes possession of the steering wheel, Australians should ask who is sitting in the driver’s seat – and, more importantly, who decides which roads we are permitted to travel.
The post Bradman read the ball. Brabham read the road. Can Canberra read AI? appeared first on The Spectator Australia.
