Rohini认为，人工智能训练数据应被视为集体管理的资源，强调社区同意、署名和利益共享。她指出，像维基百科这样的公共知识库被大型语言模型利用，但却没有维持公共领域，而仅靠版权改革是不足够的。对于全球大多数地区而言，训练数据中的代表性不足是一个关键问题，而未经社区控制的数据抓取构成了“剥削性包容”和认识论暴力。 AI
影响 呼吁新的人工智能治理模式，优先考虑训练数据的社区同意和利益共享。
排序理由 该集群包含讨论人工智能治理和数据管理权的观点文章。
在 Mastodon — sigmoid.social 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
Better # governance would treat training data as a collectively stewarded resource. That means community # consent or refusal, attribution standards, benefit-sharing, and lessons from indigenous data sovereignty. Publicly funded datasets, multilingual corpora, and open models can…
Public knowledge repositories such as # Wikipedia are commons built by volunteer labor, donations, and public-interest governance. LLMs extract value from them, but that value rarely flows back to sustain the commons. # Copyright reform alone cannot solve this. AI firms can invok…
For the Global Majority world, the dilemma is sharp. Many languages and knowledge traditions are underrepresented in training data. But “fixing” that via ungoverned scraping turns communities into data sources, not decision-makers. I argue that is 'extractive inclusion': being in…