Who Gets Seen by AI?

This week’s Field Notes suggest that algorithmic bias is no longer simply a technical flaw,

it is becoming embedded within the systems that shape modern life. Across recent reports and analyses, a consistent pattern emerges: AI systems increasingly influence access to education, employment, healthcare, and visibility, yet the data and structures behind these systems remain deeply unequal.

The Brookings analysis introduced the concept of “algorithmic exclusion,” showing that some communities are not only discriminated against by AI, but effectively erased from the datasets that power it. At the same time, UNESCO’s CULTAI report demonstrated how generative AI systems disproportionately prioritize dominant languages and Western cultural norms, leaving many African and Indigenous perspectives underrepresented. In regions such as Kenya, this raises concerns that digital inclusion without cultural representation may create a new form of invisibility.

The International AI Safety Report 2025 further revealed that bias is already producing measurable social harm in areas such as hiring, finance, and healthcare. Meanwhile, reporting from The Guardian highlighted a growing danger: societies increasingly trust AI systems because they appear objective, even when those systems inherit racial, gender, and cultural biases from the data they are trained on.

What connects these cases is a structural issue. AI does not emerge outside society; it reflects existing inequalities and can scale them globally. As automated systems become embedded into public and economic life, bias risks becoming less visible while affecting more people at once.

This reframes the debate around AI governance. The challenge is no longer simply improving algorithms, but questioning who is represented within the systems that define digital reality and who remains invisible within them.