What is the AI Ethics Assessment Tool?
The AI Ethics Assessment Tool has been developed as a part of the FAIR-AI project, which aims to develop a handbook for sustainable AI development in Austria.
The goal of the tool is to allow users the opportunity to self-assess the ethical risk of their AI-enabled projects, thereby informing themselves about the risks and implications as well as producing more responsible artefacts in the end. The tool helps with this by explaining different concepts and key terms, and suggesting actionable steps that can be taken to ensure an AI-enabled project is safe, responsible, and ethically sound.
How it works
Users rate their projects based on 7 factors of AI use and 7 factors of Ethical risk, this is done by selecting a point on a scale in response to a question for each of the 14 factors. The 7 factors of AI use explain the potential risks of each type of AI implementation, and the 7 factors of ethical risk explain how to mitigate each risk and provide actionable steps to implement in a project. Once the assessment is complete, the user can see an overview of the results, clearly showing potentially ‘problematic’ areas of their project that may require redressing or extra caution. The full report can then be generated as a PDF and downloaded, to be shared with other project stakeholders or for future reflection if the assessment is repeated at a different project stage for comparison.
Where the information comes from
All the information that is included in the tool - the factors of AI use and ethical risk, the definitions and scales, the risks, improvement suggestions and action steps - all of these come from literature and trusted academic sources. In order to come up with the content for the tool, in-depth research was done to determine what does actually define the use of AI and ethical risk. Although the assessment of a project using this tool does not equal compliance with any legal standards or requirements, these were heavily consulted in the design of the information and the EU AI Act specifically underpins a lot of the content in the Ethical Risk section.