Seth Walsh
Iconoclast
Contributor
- Joined
- Jan 12, 2020
- Posts
- 10,888
- Reputation
- 22,237
Social class was never just about salary.
That was the old lie.
Salary is what people talk about because it is visible. It is easy to compare. €45k, €80k, €150k. Nice clean numbers. People can cope with that.
But class was always about something quieter: who owns assets, who has family protection, who can take risk, who knows the rules early, who gets introduced into the right rooms, and who can survive being wrong.
AI is not going to destroy social class.
It is going to reveal it.
The old middle-class deal was simple enough. Get educated, get a respectable job, buy property, build a pension, keep your head down, become secure.
That path is breaking.
Not for everyone. For the exposed middle.
The person doing generic admin, generic analysis, generic reporting, generic marketing, generic legal drafting, generic junior coding, generic PowerPoint, generic “stakeholder management” without ownership.
That entire layer was built on information friction. Someone needed humans to move data from A to B, summarize documents, prepare decks, reconcile spreadsheets, answer basic queries, draft notes, monitor inboxes, create reports, and look busy inside large organisations.
AI attacks that directly.
IMF estimated that about 60% of jobs in advanced economies are exposed to AI, with roughly half of those potentially helped by AI and half potentially harmed through automation of key tasks. That is the class split. Same technology. Opposite outcomes.
https://www.imf.org/en/blogs/articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity
The new divide is not “white collar versus blue collar.”
It is:
Can AI make you more powerful?
Or does AI make you easier to replace?
That is the whole game.
A senior dealmaker with relationships uses AI and becomes faster.
A founder uses AI and ships with a smaller team.
A trader uses AI to research, backtest, monitor, automate.
A lawyer with clients uses AI and increases throughput.
A capital allocator uses AI to scan more opportunities.
A random analyst with no ownership uses AI and quietly trains the machine that compresses his role.
Same tool. Different class outcome.
This is why “learn to code” was incomplete advice.
The real advice was always: get close to scarce judgment, capital flows, distribution, ownership, or trusted decision-making.
Code is useful. Data is useful. AI is useful.
But if you are only a replaceable executor, the tool eats you.
If you are near decisions, assets, clients, infrastructure, risk, models, capital, or distribution, the tool compounds you.
That is why the top layer will get stronger.
Stanford’s 2025 AI Index says 78% of organizations reported using AI in 2024, up from 55% the year before, while U.S. private AI investment reached $109.1bn. This is not a side trend. It is capital reorganising labour.
https://hai.stanford.edu/ai-index/2025-ai-index-report
People think AI will make society more equal because everyone gets access to intelligence.
Wrong.
Everyone gets access to a tool.
But people do not start from the same position.
One guy uses AI from his childhood bedroom while paying no rent, with parents who know tax, property, law, and career strategy.
Another guy uses AI while renting, stressed, no capital, no fallback, no professional family advice, no network, no room for error.
Same ChatGPT subscription.
Different balance sheet.
Different nervous system.
Different outcome.
That is the brutal part.
AI gives leverage to people who already have room to use leverage.
Ireland is especially brutal for this.
Ireland is not mainly an income game. It is a balance-sheet game pretending to be an income game.
Two people can both earn €70k.
One lives at home, saves aggressively, gets deposit help, has family property knowledge, uses pension shelters, avoids rent, and can take a risky career bet.
The other pays Dublin rent from after-tax income, saves slowly, has no family capital, no inheritance expectation, no mistake buffer, and gets told to “just budget better.”
These people are not in the same class race.
They are not even on the same track.
AI makes this worse because entry-level white-collar work becomes less forgiving.
The old system allowed people to enter through low-value tasks and slowly absorb taste, norms, and judgment.
Do the spreadsheet. Sit in the meeting. Draft the note. Listen to seniors. Learn the room.
AI compresses that.
The boring junior tasks do not fully disappear, but they shrink. The remaining junior roles expect more judgment earlier. Business Insider reported recent executive survey findings that AI is making entry-level work more complex, with routine/admin tasks reduced and analytical/judgment responsibilities increasing.
https://www.businessinsider.com/entry-level-jobs-more-skilled-impact-from-ai-2026-5
This is terrible for people without early exposure.
The rich kid has already heard the language at dinner.
The professional-class kid has already seen how adults talk to lawyers, bankers, recruiters, accountants, founders.
The outsider was supposed to learn that on the job.
Now the ladder is missing rungs.
Your network matters more now, not less.
Not LinkedIn spam. Not “coffee chats.” Not fake networking events.
Real network means repeated proximity to people with agency.
People who allocate money.
People who hire.
People who build.
People who know what is fake.
People who can say one sentence about you in the right room and change your path.
AI will make generic credentials less impressive because everyone can produce polished output.
The scarce thing becomes trusted human signal.
Who vouches for you?
Who has seen you operate?
Who believes you have taste?
Who lets you near the machine?
That is class.
The fake middle class will get exposed hardest.
This is the person with a decent title, decent salary, high rent, no assets, no equity, no family help, no deep skill, no capital exposure, and no real network.
Looks successful from outside.
Fragile underneath.
They consume class signals instead of building class machinery.
Nice apartment. Nice dinners. Nice holidays. Nice clothes. No balance sheet. No ownership. No optionality.
AI does not care about the LinkedIn title.
It cares whether the work has leverage, judgment, liability, trust, or capital attached.
If not, the role gets compressed.
The real winners are not necessarily the loudest tech people.
The winners are the people who combine AI with one of these:
Capital.
Distribution.
Trust.
Regulatory permission.
Physical infrastructure.
Proprietary data.
Elite relationships.
Taste.
Risk-taking ability.
Ownership.
That is why AI infrastructure is becoming a class weapon. Data centers, chips, energy, cloud access, proprietary models, private data, workflow integration, enterprise distribution.
The public sees chatbots.
The ownership class sees compute, power, capex, contracts, margins, financing, and control.
The most important AI skill is not prompting.
Prompting is the new Excel.
Useful, not sufficient.
The real skill is knowing what should be automated, what should never be automated, where the data is dirty, where the risk sits, who owns the decision, what the incentive is, and how the workflow actually makes money.
That is why operators near messy real systems will beat clean academic prompt merchants.
A person who understands a real financial workflow, real customer pain, real operational bottlenecks, real regulatory constraints, and real capital movement can use AI seriously.
A person who only makes pretty outputs is disposable.
AI will create a new class of small-team operators.
One person with taste, domain knowledge, code, AI agents, and distribution can now do what previously needed a small department.
This is bullish for founders.
Bullish for traders.
Bullish for elite operators.
Bullish for people with unusual domain knowledge.
Bearish for generic employees hiding inside process.
The future high-status worker is not just “smart.”
He is autonomous.
He can build, sell, automate, analyse, communicate, and take responsibility.
The future low-status white-collar worker waits to be assigned tasks.
That distinction will become savage.
The biggest mistake young people will make is chasing prestige without leverage.
They will still chase titles.
Consultant.
Analyst.
Associate.
Manager.
Strategy.
Operations.
Transformation.
AI.
But most titles are packaging.
The question is simpler:
Are you getting closer to capital?
Are you getting closer to ownership?
Are you getting closer to scarce judgment?
Are you building a compounding skill?
Are you meeting serious people?
Are you accumulating assets?
Are you becoming harder to replace?
If not, it is theatre.
Class in the AI age will look calmer than people expect.
The winners will not all be flashy.
Many will look boring.
They will live below income, keep optionality, avoid rent traps, build equity, learn automation, stay close to capital flows, choose employers carefully, preserve reputation, avoid fake yield, avoid lifestyle inflation, and never let their work become generic.
The losers will not all look poor.
Many will look successful until the machine reprices them.
The conclusion is simple.
AI does not remove hierarchy.
AI accelerates hierarchy.
It widens the gap between people with leverage and people without it.
It rewards the person who can direct systems and punishes the person who is merely part of one.
The old question was:
“What job do you have?”
The new question is:
“What do you control?”
Capital, code, distribution, data, trust, relationships, taste, infrastructure, reputation, assets.
If the answer is “none,” your class position is weaker than your salary says.
If the answer compounds, your class position is stronger than your salary says.
That is the AI social class pill.
