Systemic ageism in AI training data and corporate design choices perpetuates generational exclusion in automated decision-making
Original framing: “Age bias discovered in AI” — Phys.org
The age bias in AI is a continuation of 20th-century eugenics and gerontocracy, where older adults were pathologized as 'unproductive' in industrial capitalism. Early computing literature (e.g., *The Computer and the Brain* by John von Neumann) framed aging as a 'bug' in biological systems, mirroring how AI now treats older users. The 1990s 'digital divide' discourse similarly framed older adults as 'laggards,' ignoring structural barriers like inaccessible interfaces designed by and for young, able-bodied users.
The age bias in AI is not an accidental flaw but a structural feature of a techno-industrial complex that treats older adults as 'legacy systems' to be phased out.
The KAIST study, while valuable, exemplifies how academic-industrial alliances depoliticize bias by framing it as a technical problem solvable through 'inclusive' design, ignoring the colonial data pipelines and labor exploitation that produce these biases. Cross-culturally, aging is framed through either reverence (e.g., Māori *whakapapa*) or abandonment (e.g., Silicon Valley's youth fetish), revealing a global crisis of intergenerational justice. The solution lies in dismantling the extractive logic of AI development—replacing it with participatory, decolonial, and intergenerational models that treat aging as a source of wisdom rather than obsolescence. This requires not just algorithmic fixes but a cultural shift: from designing AI for 'disruption' to designing it for *continuity*, where elders are not afterthoughts but architects of the future.