For several generations there was an unwritten bargain in Western society. If you went to university, acquired the appropriate credentials and entered one of the respectable professions, you were supposed to be largely insulated from the economic storms that battered everyone else. Factories could close, mines could shut, manufacturing could move to China, a machine could replace ten men on a production line, the local shopping strip could be destroyed by a multinational retailer and the truck driver could eventually face competition from an autonomous vehicle. These were regrettable consequences of progress, we were told. The educated professional, meanwhile, sat comfortably above the battlefield.
That comfortable assumption may now be ending. Maximillian Garely, writing in The Spectator, calls artificial intelligence a form of "cosmic justice" for the professional classes. The phrase is deliberately provocative, but the irony behind it is difficult to miss. For decades, technological disruption was something that happened mainly to other people. Now the technology advancing most rapidly is unusually good at doing precisely the sort of things educated people are paid to do.
It writes, summarises, calculates and searches enormous bodies of information. It drafts contracts, analyses financial documents, produces computer code, prepares reports, translates languages and examines medical information. It can consume in seconds material that would take a junior professional days to read. The bulldozer has finally arrived at the office.
There is no reason to gloat about this. A lawyer with a mortgage and children who loses his job experiences unemployment no less painfully than a machinist. The accountant who discovers at 52 that software can perform much of what once justified his salary has every right to be frightened. The young graduate who accumulated an enormous university debt on the promise that professional qualifications would guarantee a secure future has legitimate grounds for feeling betrayed. But the historical irony remains.
Industrial workers have been living with this insecurity for half a century. From the 1970s onwards, Western manufacturing communities were repeatedly told that economic transformation was unavoidable. Factories were automated or relocated overseas, tariffs came down, supply chains became global, employers discovered cheaper labour elsewhere and entire towns could lose the industries around which they had been constructed.
The economic argument was often impeccable. Consumers received cheaper goods, companies became more efficient, capital flowed towards more productive uses and economists could demonstrate that society as a whole became wealthier. None of that was much consolation to the 55-year-old factory worker whose job had disappeared.
The response of the educated classes was frequently less than sympathetic. Workers resisting globalisation were protectionists. People worried about their jobs were afraid of progress. Communities objecting to factory closures needed to adapt, while workers displaced by technological change should acquire new skills. Eventually the insult was compressed into three words: learn to code.
Now the coder may need to learn something else. Garely points to a remarkable change in American attitudes. Gallup found that 27 per cent of US employees now worry that technology will make their jobs obsolete, more than double the proportion recorded when the question was first asked in 2017. More significantly, the old educational divide in that anxiety has almost disappeared. College graduates are increasingly frightened of technological displacement too.
The professional classes are discovering an unpleasant economic truth that blue-collar workers learned long ago: education does not repeal creative destruction. Australia is unlikely to escape it. Deloitte Access Economics recently identified 82 occupations particularly exposed to disruption as AI replaces tasks requiring relatively little human judgment, empathy or interpersonal interaction. Its assessment is that white-collar employment faces particularly significant disruption as artificial intelligence moves further into the Australian workplace.
Law provides an almost perfect example. For generations the legal profession operated behind formidable barriers to entry. One needed a degree, admission, specialised knowledge and access to expensive databases and libraries. Much of the profession's economic value arose because ordinary people simply could not navigate the machinery themselves. AI changes part of that equation.
AI can already summarise cases, locate authorities, compare contracts, prepare chronologies, analyse discovery and produce first drafts of legal documents, all with errors. In September, Reuters reported the launch of FairPlay Law, an AI-native American employment law firm using artificial intelligence to analyse workplace documents and offering flat fees rather than relying entirely upon conventional hourly billing. At the upper end of the profession the movement is equally obvious, with major international law firms investing heavily in specialised legal AI.
The immediate future probably does not involve replacing the senior barrister conducting a difficult cross-examination. It may involve needing far fewer junior people to perform the mountain of preparatory work beneath him. That distinction is crucial because the first great effect of AI may not be mass unemployment. It may instead be the hollowing out of professional labour.
Suppose a law firm previously needed ten junior lawyers and AI enables five to produce the same output. Five jobs have disappeared even though the profession of law survives. Suppose an accounting practice can service twice as many clients with the same number of accountants, or one advertising employee can generate what previously required a small creative department. Nothing needs to become completely automated for labour markets to be transformed.
Medicine presents a more complicated picture because the physical patient remains stubbornly human. People want a doctor to examine them, make difficult judgments and take responsibility. But considerable amounts of medical work involve pattern recognition, documentation, administrative processing and the interpretation of information. AI does not have to replace the doctor to change radically how many people are required around the doctor.
The same applies across the professional economy. Accountants, financial analysts, programmers, architects, consultants, academics, journalists, administrators, paralegals, researchers, public servants and middle managers perform varying mixtures of tasks. AI does not need to master every task within an occupation. It merely needs to automate enough valuable tasks to alter the economics of employing people to perform them.
There is, of course, another possible future. AI could make professionals vastly more productive without substantially reducing their numbers. Lower costs might generate greater demand. Lawyers could become affordable to people who currently cannot afford legal representation, doctors might spend less time completing paperwork and more time treating patients, researchers could investigate questions previously impossible because of the labour required, and small businesses could obtain analytical capabilities once available only to large corporations.
PwC's 2026 research points towards precisely this complexity. It distinguishes occupations in which AI reduces the expertise required from those in which automation of routine work leaves humans concentrating upon more sophisticated tasks. Australian advertisements for the latter category have actually been growing strongly. Technology has repeatedly created occupations that previous generations could scarcely have imagined, so predictions of the imminent extinction of the professional classes should be treated with the same scepticism as every other confident prediction about AI.
But something has already changed psychologically: the professional class can no longer assume immunity. That may have political consequences extending far beyond employment statistics.
For decades the people who administered globalisation were disproportionately insulated from its harshest consequences. Economists recommended structural adjustment, consultants designed efficiencies, lawyers constructed the agreements, academics explained why resistance was irrational and journalists described displaced industrial communities as places that needed to adapt. The professional-managerial class generally remained employed.
This produced one of the great social asymmetries of the globalised era. The people most enthusiastic about economic disruption were often not the people most exposed to it. AI potentially changes that arrangement.
Imagine telling a redundant solicitor that artificial intelligence will increase aggregate productivity. Tell the unemployed accountant that society as a whole benefits from cheaper financial services. Explain to the displaced journalist that technological progress cannot be stopped. Advise the 48-year-old middle manager to retrain for the industries of tomorrow. Perhaps tell him to learn to prompt.
The joke is cruel, which is precisely why it exposes the earlier cruelty. People whose livelihoods disappear do not experience economic change as a graph showing productivity growth. They experience it as fear: mortgage payments, school fees, lost status, marital strain, depleted savings and the humiliating business of discovering that a society which yesterday regarded their skills as valuable no longer requires them. That was true of the factory worker, and it will be equally true of the lawyer.
The appropriate response therefore isn't Schadenfreude. It is solidarity and perhaps a little humility. If artificial intelligence does seriously disrupt professional employment, the chattering classes may discover that many of the people they once dismissed as economically backward were describing a genuine human problem. Employment is more than income. Occupations provide identity, status, routine, community and the satisfaction of believing that one's abilities are useful to other people. Losing that cannot always be repaired by a government retraining program.
There is a further irony. The professional classes may prove much more politically effective in defending themselves than the industrial workers were. If AI begins eliminating significant numbers of lawyers, academics, journalists, administrators and other credentialled professionals, we may suddenly hear sophisticated arguments explaining why unrestrained technological competition is socially destructive.
Parliamentary committees may appear, ethical principles will be discovered, professional associations will demand safeguards and newspapers may develop a remarkable new sensitivity to the social costs of automation. Perhaps all of those things will be justified. But the factory worker could reasonably ask where this wisdom was when his factory closed.
That is the real meaning of the "cosmic justice" argument. It is not that professionals deserve to suffer. Nobody deserves to lose a livelihood merely so somebody else can enjoy the symmetry. The justice, if there is any, lies in the destruction of an illusion.
Economic insecurity was never a peculiar defect of working-class life. It was something from which the credentialled classes happened, for a historical period, to be unusually well protected. Artificial intelligence may remove that protection.
If it does, the best outcome would not be that lawyers learn what it feels like to be factory workers. It would be that lawyers, academics, journalists and managers finally understand why the factory workers were angry in the first place.
https://spectator.com/article/ai-is-cosmic-justice-for-the-professional-classes/