AI and Opportunity — Distribution or Concentration?

Every major technology wave in the last century has been accompanied by claims that it will democratise opportunity — that it will break down the barriers that concentrate economic power in the hands of the few and distribute it more broadly. The printing press. Electricity. The internet. Now AI.

The track record is mixed. Each technology has created new opportunities for some people and eliminated opportunities for others. The distribution of who benefits and who loses has rarely matched the optimistic prediction.

The question is not whether AI will democratise opportunity — that framing is too simple for what is actually happening. The question is what AI concentrates, what it distributes, and who is positioned to capture which effect.

What AI is distributing

Capability once reserved for experts. The ability to draft a contract, to analyse a dataset, to debug code, to design a marketing campaign, to write a performance review — these tasks previously required either specialist training or expensive access to specialists. AI tools make them accessible to anyone who can use the tools. This is genuine distribution of capability that was previously concentrated.

The ability to produce at a higher level. A writer with AI assistance can produce better-edited, better-structured work than the same writer without it. A programmer with AI assistance can debug faster and write more complete code. A researcher with AI assistance can synthesise more information more quickly. The quality ceiling for what any given person can produce has risen. This benefits people throughout the capability distribution, not just those at the top.

Access to information and explanation. The ability to ask any question and receive a coherent, expert-level answer — and to follow up until the answer makes sense — is distributed universally to anyone with an internet connection. For people who grew up in environments where access to expert knowledge was scarce or expensive, this is a genuinely significant change.

What AI is concentrating

The returns to capital. AI infrastructure is expensive. The compute, the data, and the engineering talent required to build and operate frontier AI systems are concentrated in a small number of well-capitalised companies. The economic returns to AI capability accrue disproportionately to the owners of that infrastructure. The distribution of AI tools to users does not distribute the economic returns to the infrastructure that produces those tools.

The productivity advantage of the already-capable. AI tools amplify existing capability more than they substitute for absent capability. The experienced programmer who uses AI to write code faster produces more in a day than the novice programmer who relies on AI to write code they don't understand. The experienced writer who uses AI for editing produces better work faster than the novice writer who uses AI to write for them. The productivity gains from AI are not equally distributed — they are larger for people who have more underlying capability to amplify.

The advantage of capital-rich markets. AI tools are improving fastest in English-language, US-centric contexts. The models are trained primarily on English-language data. The products are built for markets with high purchasing power. The benefits of the best AI tools are most accessible to people in the most developed markets. This is a concentration of advantage that runs counter to the democratisation narrative.

The honest picture

Both things are happening simultaneously, and the balance between them depends on specific context.

In the specific context of someone in a low-income country with internet access and the motivation to develop technical skills: AI is genuinely distributing opportunity. The ability to learn programming from AI tutoring tools, to build software with AI assistance, and to sell that software or those services to global markets is a path that is more accessible now than it was five years ago.

In the specific context of the competition for the most economically valuable opportunities — senior leadership roles, ownership stakes in high-growth companies, positions in the most productive institutions — AI is doing relatively little to change who wins. The selection systems that gate these opportunities are built on credentials, networks, and demonstrated track records that AI does not produce.

The distribution of opportunity is happening at the level of capability. The concentration of opportunity is happening at the level of capital returns and access to the highest-value positions. Both are real. The people who are best positioned to benefit from the distribution while protecting themselves from the concentration are those who are developing genuine AI-amplified capabilities and positioning those capabilities where global demand is strongest.

That path is available from more places than it used to be. It requires more deliberate navigation than the optimistic case suggests.