LNAT model essay: AI and inequality

It separates two questions most answers run together, jobs and inequality, and that distinction carries the argument. 524 words.

1. Automation and artificial intelligence are likely to make society more unequal unless governments act to share their gains. The technology itself does not decide who wins. Left alone, however, it rewards those who own machines and those whose skills machines cannot copy, and the rest of the workforce competes for what remains.

   NoteA clear prediction with its condition. The writer also says what the claim is not: technology is not destiny.

2. The main reason is that automation changes who earns the returns. When software can do a task, the company that owns the software keeps the money a worker used to earn. The benefits go to shareholders and to a smaller group of highly skilled staff who design and manage the systems. Routine office and factory work, which once offered steady middle incomes, is exactly the work machines are best at. Even where jobs remain, workers who could be replaced by software lose bargaining power, and their pay falls with it.

   NoteA mechanism, not a slogan. The paragraph explains how inequality would rise, which is more persuasive than saying that it will.

3. The gains are also likely to be concentrated in places as well as in people. The firms that build the most advanced systems are few, large and based in a small number of cities and countries. Regions whose economies rest on work that can be automated, such as call centres or warehouses, may lose jobs without gaining the new ones, which tend to cluster where the companies and skilled workers already are. Inequality between places can be as damaging as inequality between individuals.

   NoteA second reason that extends the argument from individuals to regions, so the essay is not resting on one mechanism alone.

4. The best objection comes from history. Every wave of new technology, from the spinning jenny to the personal computer, was expected to destroy work, and each eventually created more jobs than it removed. Most people today do jobs that did not exist a century ago. On this view, fears about AI repeat an old mistake. Economists who hold this view point out that rising productivity tends to make goods cheaper, which raises living standards even for those whose wages do not rise.

   NoteThe objection is strong and drawn from evidence the reader will recognise. It is not a straw man.

5. But the number of jobs and the spread of income are different questions. A country can have plenty of work and still grow more unequal, if the new jobs are either highly paid or insecure and poorly paid. The history the objection relies on also includes long, painful transitions: workers displaced in one decade did not usually walk into the new jobs of the next. Many of the hand-loom weavers replaced by machines in the nineteenth century never found work as good again; the gains went to their grandchildren. Whether AI widens inequality therefore depends on training, tax and welfare policy, not on the technology alone. Cheaper goods help, but they do not close the gap between those who own the new systems and those who are replaced by them.

   NoteThe reply finds the gap in the objection: it answers a question about jobs, not inequality. The final sentence returns to the condition set in the opening.

6. So automation is not bound to divide society, but it will if nothing is done. Retraining that reaches workers before their jobs disappear, and taxes that capture some of the gains from automation, could spread the benefits widely. Governments have managed such shifts before: free schooling, unemployment insurance and pensions were all responses to earlier changes in how work was organised. The choice belongs to governments, and the default, if they make none, is a more unequal society. That is a reason for urgency, not for despair.

   NoteThe ending names concrete policy levers and restates the answer, so a reader who skims to the end still sees the position.

What would make it stronger: Add one specific example of a job that automation has already changed, such as checkout staff or call centres, to make the mechanism concrete.