02三项职责
从三个角度看该领域。
03系统性问题
From model failure to systemic effect.
Individual AI failures (biased outputs, hallucinated facts, unsafe actions by an agent) are well catalogued. Systemic effects arise when many actors use similar models trained on similar data, when AI is embedded in critical decisions such as credit or triage, or when capability shifts faster than oversight. Assessment has to cover the model, the system it sits in and the institution operating it.
| 图层 | 典型风险 | 评估 |
|---|---|---|
| 建模 | Robustness, bias, capability for misuse | Evaluations and red teaming |
| 系统 | Integration errors, over-reliance, data leakage | 影响评估、 人工监督 设计 |
| 机构 | Unclear accountability, missing incident response | Management system audit (ISO/IEC 42001) |
| 生态系统 | Homogeneity, concentration, correlated failure | 市场和依赖关系分析 |
04框架
风险框架与法律。
| 框架或机构 | 其管辖或提供的内容 |
|---|---|
| NIST AI RMF 及其生成式人工智能概况 | 治理、映射、衡量和管理四项职能 |
| ISO/IEC 42001 和 ISO/IEC 23894 | AI management systems and AI risk management guidance |
| 欧盟《人工智能法》 | 按风险分级的义务分阶段实施;请核对各项义务的现行日期 |
| 经合组织人工智能原则和联合国教科文组织《人工智能伦理问题建议书》 | Intergovernmental principles |
07来源
Where this page’s references lead.
阅读本页时所对照的机构和文件。状态和版本会变化;每个链接都指向发布机构。
08问题
问题与解答。
GCRI 为何托管现场网关?
GCRI hosts the shared references of the Consortium, including the STEM Commons and the field gateways, so each field has one place that links all three institutions’ work.
报告贡献
从本页贡献证据。
Reports 是联合体的季刊。您对该主题的了解可以成为下一期中的证据、案例或更正,署名归您,并在属于您本人作品时以 CC BY 4.0 发布。
GCRI
- 电子邮件网关 GW-GCRI从本页发送证据、更正、案例、方法、审查、翻译或概况投稿。它会带上本页标签送达,由负责的编辑接收。
- ZenodoOpen depositDeposit it and submit to the communityPublish your own record on Zenodo with your ORCID iD, then submit it to the Reports community there. Curators review it; the DOI stays yours.
- 报告GCRI查看所有贡献方式贡献类型、各自的评审方式、您获得的署名、许可协议,以及绝不接受的内容。
切勿发送个人数据、敏感基础设施细节或您无权分享的任何内容。贡献在使用前会经过审查;发送贡献不赋予署名权或编辑权,赞助也绝不能换取研究结论。
法律
法律文件。
本网站上的地图、数据和图表仅作为背景和一般信息展示。它们不是建议、预测或预警;信息的局限见免责声明和数据说明。