AI and Equity

The benefits and risks of AI are not distributed equally across populations. Understanding who benefits and who bears risk is essential to thinking about AI ethically.

The global distribution

The most capable AI systems are developed by a small number of wealthy companies, mostly in the US and China. The economic benefits of AI largely accrue to the shareholders and employees of these companies, and to organisations wealthy enough to deploy the technology effectively. Meanwhile, the labour that trained many AI systems — rating outputs, labelling data — is often done by low-wage workers in Kenya, the Philippines, and elsewhere.

Intra-country inequality

Within countries like New Zealand, AI productivity gains may largely flow to capital rather than labour, increasing inequality. Workers displaced by AI may not easily transition into the new jobs AI creates, particularly if those jobs require different skills or are in different locations.

Representation in the data

AI systems work better for groups that are well-represented in training data. English-speaking, Western, formally educated individuals are well-represented. Speakers of minority languages, people with non-mainstream cultural contexts, and individuals whose concerns are underrepresented in digital data are at a systematic disadvantage.

What you can do

Paying attention to who bears the risks and who gets the rewards from AI systems — and advocating for more equitable distribution — is a form of civic engagement that your generation will need to engage with throughout your lives.

AI does not automatically make the world more equal or less equal — that depends on the policy choices societies make about how AI is governed, taxed, and regulated. Those choices are political, which means they can be contested.