AI as the Great Equaliser

Artificial intelligence has a characteristic that sets it apart from many earlier technological revolutions: it does not seem destined to transform only particular occupations or departments, but to cut across almost the entire structure of an organisation at the same time. It affects the administrator preparing documents, the designer, the programmer, the marketing department, the accountant, the lawyer, the chief financial officer and the chief executive too. The tasks change, but no one is really left out.

That is why I like to think of artificial intelligence as a great equaliser. Not because it will equalise everyone’s salaries, abilities or opportunities. The exact opposite will probably happen, with enormous differences emerging between those who learn to use it well and those who do not. The equalisation happens in another sense: a person’s position within an organisation ceases to protect them from technological change.

We can imagine a company as a building by the sea. On the lower floors are administrative and operational jobs; higher up, specialists and department heads; then management; and on the top floor, the CEO. For a long time, it was reasonable to think that a major technological transformation would enter from below. First it would automate simple, repetitive tasks, and only much later would it reach jobs requiring experience, judgement or authority.

This time, the wave is taller than the building.

The administrator may discover that a substantial part of their work can be automated. But the senior lawyer may discover at the same time that a system researches, compares documents and prepares a first draft at a speed impossible for a person. A marketing manager may find that one person equipped with good tools produces what, until recently, required a small team. A programmer begins working with agents capable of spending hours carrying out entire tasks. And a CEO may have to decide how to transform their company using a technology they understand considerably less well than some employees many levels below them.

The position still matters, of course. So do experience and knowledge. But none of them provides shelter from the wave.

There is also a fairly obvious paradox: never has there been so much talk about artificial intelligence, and at the same time, probably never have so many people expressed opinions about a technology they understand so little.

Under the term “artificial intelligence”, we place very different things: language models, chatbots, vision systems, recommendation algorithms, agents capable of using tools, models running locally, networks of agents cooperating with one another. For a large part of the public, all of this ends up being roughly the same thing. And when we do not understand a technology well, we tend to fill in the unknown using human metaphors.

That is why we sometimes hear that the model “thinks”, “wants”, “is afraid” or “tries to escape”. Even people in exceptionally influential positions, researchers and former technology executives have warned of scenarios involving loss of control or existential threats, sometimes using language more reminiscent of Terminator than of everyday technical analysis.

The discussion of extreme risks is legitimate, although those may not be the risks.

To understand the economic transformation already before us, we do not need to settle whether a machine might one day become conscious. An artificial intelligence does not need consciousness, desires of its own or ambition to radically change the work of a lawyer, an accountant, a designer or an executive. It only needs to be useful.

And it already is.

It can read quantities of documentation that a person could not process in a reasonable time, find relationships between texts, write code, translate, analyse images, operate tools, prepare alternatives, break a complex problem into tasks and delegate them to other systems. Each of these capabilities is open to debate, involves errors and needs controls. But their combination is already changing the cost of intellectual work.

That is probably the fundamental economic change. Artificial intelligence is making cognitive capacity cheaper in much the same way that computing made calculation cheaper.

This does not mean that human knowledge loses its value. Rather, it changes where that value lies. When producing a first draft, searching for information or generating a hundred alternatives becomes very cheap, deciding which problem is worth solving, which information is relevant, how the process is organised, which result is correct and who is willing to answer for it becomes more important.

The consequence is especially interesting for anyone without a large organisation behind them.

Until very recently, many capabilities were tied to size. Researching a market, developing software, preparing campaigns, analysing large quantities of documents or producing content at scale required specialist teams or hiring different suppliers. Today, through artificial intelligence, a single person can mobilise an amount of intellectual capacity that would have required a substantial organisation just a few years ago.

That does not magically turn anyone into a good business owner, a good lawyer or a good designer. Access to an excavator does not make someone an engineer. But it substantially changes what a competent person can do with the resources available.

And here another dimension of the great equalisation emerges: for a brief period, we have all become beginners again.

A chief executive with thirty years of experience and an employee who joined six months ago may be discovering the same tool in the same week. The former has vastly more knowledge of the company, the market and its people. That advantage remains enormous. But the latter may spend hundreds of hours experimenting with new models, agents and tools while the executive remains convinced that AI is fundamentally about typing questions into ChatGPT.

In two years, the technological distance between them may be very significant.

In fact, the higher a person’s position, the greater their interest in understanding this technology should be. An executive does not need to become a model engineer, but will have to decide which systems their company buys, which processes it automates, which information is given to those systems, which risks it accepts and which employees are ready to use them. It is difficult to govern something without at least a basic understanding of it.

That is why AI literacy is beginning to look to me like a professional skill needed across all roles, comparable in some ways to what happened with computing or the internet, though probably at a much greater speed.

And literacy does not mean learning five tricks for writing better prompts.

That was an initial stage. When interaction consisted mainly of talking to a chatbot, it made sense to concentrate on how to formulate a good request. But today’s systems can already incorporate memory, tools, browsing, code execution and specialist agents. Even a basic understanding of what we are using requires knowing the difference between a model and an agent, what a context window is, why tokens matter, what happens when we grant access to a tool, which information may leave our organisation, which results need verification and which actions should require authorisation.

You do not need to understand how an engine is manufactured to drive a car. But it does help to know what the brake, the accelerator and the dashboard indicators are for.

With artificial intelligence, we are still learning even where those controls are.

The natural reaction to all of this is fear. If a machine can do part of my work, the first question is almost inevitable: will it replace me?

There is no single answer. Some roles will disappear. Others will need far fewer people. Many will be transformed, and activities will emerge that we do not yet know how to name. It is difficult to imagine a transformation of this magnitude without creative destruction, winners and losers.

But focusing solely on that question puts the professional in a strangely passive position. There is another that is considerably more useful: what can I do now with this technology that I could not do before?

That question changes the perspective.

For an administrator, it may mean eliminating hours of repetitive tasks. For a lawyer, analysing and comparing a quantity of documentation that would previously have taken several days. For a small business owner, having research, programming, design and analytical capabilities without immediately hiring a department for each. For an executive, exploring scenarios and preparing decisions in a depth that previously required bringing several people together.

AI can destroy certain advantages, but it also creates new ones.

That is why describing it as a great equaliser does not mean predicting a more egalitarian society. It may produce exactly the opposite. What it does is redistribute the sources of advantage.

The size of the organisation may carry somewhat less weight. Access to generic knowledge may be worth less. The ability to produce many pages, many lines of code or many presentations will probably be worth less. By contrast, judgement and the ability to define problems, integrate knowledge, evaluate results and organise systems may be worth much more.

Two people who hold exactly the same position today may find themselves in completely different professional circumstances a few years from now, depending on how they have incorporated these tools into their work.

And someone relatively low in a hierarchy may come to possess an intellectual production capacity comparable to what previously required a department.

That is the genuinely equalising part.

For a long time, we have spoken about artificial intelligence in the future tense: when it arrives, when it becomes capable, when companies begin to use it.

That language is ageing quickly.

The wave has already arrived.

We do not know exactly how far it will reach, or how much of today’s enthusiasm will prove exaggerated. But it seems increasingly unlikely that intellectual work will be organised in five or ten years as it is today.

Faced with a wave of this size, you can learn to swim, try to harness its power to sail, or simply hope it does not reach you. The problem is that, however high the floor from which you watch, the wave is still taller than the building.

Artificial intelligence does not ask for our job title before entering.

And perhaps that is its most profoundly equalising quality: it forces all of us, from the administrator to the chief executive, to learn again.