FairCVreview: A novel platform to advance trust in the compliance and fairness of AI systems for the recruitment sector by providing fully transparent and bias-free AI systems.

FairCVreview：一个新颖的平台，通过提供完全透明和无偏见的人工智能系统，增强对招聘领域人工智能系统合规性和公平性的信任。

• 批准号: 10076196
• 金额: $ 6.33万
• 依托单位: NEUROMIND LABS LTD
• 依托单位国家: 英国
• 项目类别: Grant for R&D
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=10076196
• 关键词: FairCVreview; novel; platform; advance; trust

FairCVreview is an AI-based CV screening tool that focuses on fairness and transparency. It is based on an easy-to-use interface that gives even the smallest recruitment companies access to the latest technologies without needing coding or technical skills to operate.The screening technology focuses on fairness, using AI to identify and eliminate the bias in underlying data. The outputs are fully explained so that recruiters, job applicants and employers can all have full confidence in the fairness of the selection.Recruiters and employers will see details of the decision-making process and get metrics on equality parameters. They will be able to share details with applicants to explain why they have not been selected (e.g. indicating which skills or qualifications were lacking compared to the selected candidates).Diversity metrics are provided so employers can identify any issues and take steps to address them. For example, if the candidate pool lacks diversity then the employer could improve their advertising strategies to encourage a broader range of applicants.Additionally, Job seekers will be able to test their own CV via our additional public-facing application. This will provide a user-friendly interface allowing individuals to upload their CV and it will flag any elements that could potentially expose them to bias. For example, if they have included details that give away their age, they will be prompted to consider if this is necessary.FairCVreview will be developed using a collaborative approach with input from customer representatives (recruiters).Firstly, we will identify the types of bias that could potentially impact the project and we will identify the target groups and key dimensions that could be impacted (e.g. gender, ethnicity, age, etc).This includes working with subject experts to identify bias that exists in the underlying data as well as any that can arise during machine learning processing. During AI development and training, we will evaluate the outputs for bias across all potential dimensions and these will be scored against performance metrics, allowing us to identify and mitigate against any bias.Secondly, we will deliver transparent results, giving feedback to the recruiter/employer on exactly why the CV achieved that match score. To achieve this, we will build on our experience in developing similar explainability technology for the essay marking sector (which uses a similar approach to eliminate bias and deliver transparent results).

Faircvreview是一种基于AI的CV筛选工具，侧重于公平性和透明度。它基于一个易于使用的界面，即使是最小的招聘公司也可以访问最新技术，而无需进行编码或技术技能即可操作。筛选技术专注于公平性，使用AI来识别和消除基础数据中的偏见。产出的解释得到了充分的解释，以便招聘人员，求职者和雇主都可以对选择的公平有充分的信心。雇主和雇主将看到决策过程的详细信息，并获得有关平等参数的指标。他们将能够与申请人共享详细信息，以解释为什么没有选择他们（例如，与选定的候选人相比，指示缺乏哪些技能或资格）。提供多样性指标，以便雇主可以识别任何问题并采取措施解决这些问题。例如，如果候选人池缺乏多样性，那么雇主可以改善他们的广告策略，以鼓励更广泛的申请人。在此方面，求职者将能够通过我们的额外公开申请来测试自己的简历。这将提供一个用户友好的接口，允许个人上传其简历，它将标记任何可能使其偏见的元素。 For example, if they have included details that give away their age, they will be prompted to consider if this is necessary.FairCVreview will be developed using a collaborative approach with input from customer representatives (recruiters).Firstly, we will identify the types of bias that could potentially impact the project and we will identify the target groups and key dimensions that could be impacted (e.g. gender, ethnicity, age, etc).This includes working with subject experts to identify基础数据以及机器学习处理过程中可能出现的任何偏见。在AI开发和培训期间，我们将评估所有潜在维度的偏差的输出，这些输出将与性能指标进行评分，从而使我们能够识别和减轻任何BIA的识别和减轻任何BIA。首先，我们将提供透明的结果，从而给招聘人员/雇主提供反馈，从而确切地对CV取得了匹配得分。为了实现这一目标，我们将在为论文标记部门开发类似的解释性技术方面的经验（使用类似的方法来消除偏见并提供透明的结果）。