AI in the middle class · v1.1.0

Why the middle class hate AI

A compact editorial app: the first page shows the flow from conversation to method to projection, and the second page holds the final FT-style article.

Page 1

From discussion to article

The process page keeps the source method visible without letting its internal terminology leak into the article. The slide matters because it gave us the rule for the whole argument: judge AI by what its outputs make people and institutions do, not by whether it looks like human consciousness.

01

Discussion

Start with the public reaction to leaked AI financials and the wider claim that educated critics often read every fact as evidence against AI.

02

Method layer

Use the slide as the hidden analytical engine. It says intelligence can be understood functionally: inputs are transformed into outputs that change decisions, work, coordination and control. This is necessary because it stops the article becoming a metaphysical argument about whether AI “really understands”.

Method slide showing token input, transformation and output as a functional account of intelligence.
Source method retained for the process page, deliberately excluded from the final article language.
03

Manipulation

Convert the method into editorial claims: AI matters because it turns rough material into usable work, and that threatens the status of people whose authority rests on producing polished judgement, language and procedure.

04

Projection

Project the claim into an FT-style international article: short sentences, controlled tone, no jargon, no slide terminology, clear concessions, and a compressed argument that can travel across markets and cultures.

05

Result

The final article keeps the functional insight but hides the scaffolding. It says AI is hated by parts of the middle class because it cheapens the symbolic work that once protected professional status.

Page 2

Why the middle class hate AI

The fiercest resistance to artificial intelligence does not come from those with the least education. It comes from those whose education once guaranteed control over words, judgment and institutional process. AI is therefore not only a technology story. It is a class story.

The middle class built much of its authority on symbolic work. It wrote the memo, marked the exam, interpreted the rule, drafted the contract, produced the slide deck, translated the data, summarised the meeting and explained the world to everyone else. These tasks were not merely jobs. They were status signals. They showed who had been trained, selected and admitted.

AI attacks that settlement directly. Its most disruptive feature is not that it thinks like a person. It is that it turns rough input into usable output at scale. A prompt becomes a letter. A document becomes a summary. A question becomes an answer. A dataset becomes a recommendation. A vague intention becomes a first draft of action. This is precisely the layer of work on which many professional identities depend.

That is why so much criticism of AI has a restless quality. The objection keeps moving. It is said to be useless, then too powerful. It is said to be fake, then dangerous. It is condemned for making mistakes, then condemned for replacing people despite those mistakes. It is attacked as theft, pollution, hype, monopoly, mediocrity and fraud. Some of these criticisms contain truth. Taken together, however, they often reveal less an argument than a wish: that the technology will somehow be stopped before it changes the terms of professional life.

The recent focus on AI companies’ losses fits this pattern. High spending and heavy losses may be a serious business problem. They are not proof that the product has no value. Railways, electricity, telecoms, semiconductors, cloud computing and the internet all passed through periods of waste, speculation and overbuild. The fact that investors may overpay for infrastructure does not mean the infrastructure is unimportant. Often it means the opposite.

The relevant question is not whether AI is pure, conscious or polite enough for the seminar room. It is whether it changes behaviour. It plainly does. Workers use it to code, search, write, design, teach, analyse and decide. Companies use it to reduce friction in processes that once required layers of junior professional labour. Consumers use it because it gives them access to forms of fluency that were previously scarce. That is not a philosophical claim. It is an operational one.

This is why the middle-class critique can sound so moral while being so material. AI lowers the price of competence in areas where competence was once protected by credentials. It weakens the premium on being able to produce polished language, plausible analysis and administrative order. It does not abolish expertise. But it changes the boundary between expert and non-expert, and that is enough to provoke hostility from people whose position depends on that boundary.

The better response is not denial. AI will be overhyped. It will be misused. It will produce errors. It will concentrate power if left entirely to the largest platforms. It will also raise productivity, broaden access to useful skills and force many institutions to explain what their credentials are really worth. The task is to govern the transition without pretending the transition is not happening.

The middle class hates AI because it recognises the threat before it can fully name it. The machine is not coming first for manual labour. It is coming for the routine production of judgment, language and procedure. That is uncomfortable. It is also why the technology matters.