Merge branch 'main' into dev

Author: incd-github-actions[bot]

src/collections/risks/en/r13.md

@@ -1,6 +1,7 @@
 ---
 index: R13
 title: Exploitation of under-employed persons with disabilities in AI gig work
+shortTitle: Exploitation
 excerpt: Low-paid AI data work may exploit people with disabilities in ways that reinforce economic inequality and risk worsening health conditions.
 risk: Low-paid data labelling, reinforcement learning (from human feedback), and content-moderation tasks—the "ghost work" that trains AI systems—disproportionately falls on under-employed persons with disabilities in low- and middle-income countries (LMICs), often under conditions that exacerbate disability. This pattern is documented in academia in decolonial AI literature as a defining feature of algorithmic coloniality.
 mitigation: Living-wage floors should be required for AI data work, and accessibility audits should apply to gig platforms themselves, as opposed to being applied only to the AI products built on top of them. Recognition of labelling labour in AI supply-chain reporting, a right to disconnect, and medical and ergonomic protections should be established.

src/collections/risks/en/r14.md

@@ -1,6 +1,7 @@
 ---
 index: R14
 title: Evaluation and mitigation of harm using statistical approaches
+shortTitle: Bias in evaluation
 excerpt: Harm against persons with disabilities is determined to be anecdotal or statistically insignificant when using statistical determination of negative impact or risk. Mitigation strategies that rely on statistical reasoning are subject to bias against statistical outliers.
 risk: When harm is assessed using statistical approaches, impacts on people with disabilities may be dismissed as anecdotal or not statistically significant enough to act on. As a result, mitigation approaches that depend on these methods may be more likely to overlook or fail to address harms affecting people whose experiences fall outside common patterns.
 mitigation: Collect and analyze individual reports of harm alongside statistical data, using tools such as an incident database. This approach creates a structured way to document real-world experiences, including detailed descriptions of what happened, who was affected, and what the impact was. Including anecdotal evidence helps ensure that lived experiences are treated as valid forms of evidence, not just as isolated incidents. It can also support earlier detection of harm, especially in situations where waiting for large amounts of data could delay action.