XAI-Canvas - Handout for the human-centered design process
In the field of artificial intelligence (AI), the explainability of AI decisions is becoming increasingly important. With the XAI canvas, KARL provides an overview of relevant aspects of the use of explainable artificial intelligence (XAI). What benefit should be achieved by explaining the AI decisions? To whom should the AI decisions be explained and what requirements do these people have for a helpful explanation?
Target group
The XAI canvas is aimed at all people and companies who want to add explainability components to an AI application in order to better understand AI decisions.
Expected result
The XAI canvas provides support for the initial phases of problem analysis according to the human-centered design process (DIN EN ISO 9241-210): Understanding the context of use and defining user requirements. The XAI canvas supports companies in defining the initial vision for the use of XAI in an AI application. Filling out the XAI canvas first defines the overarching objectives of the declarations and the benefits that users expect from them. The canvas is also intended to raise awareness of any legal obligations that could make the use of XAI necessary. Initial XAI user stories are then created. These provide an initial overview of the different users and their requirements for helpful explanations.
General conditions
It is assumed that an overarching AI use case has already been defined in the company, as this provides the basis for the XAI aspects under consideration. If an AI use case does not yet exist, the KARL AI selection system can help to identify a suitable AI use case for companies. Furthermore, the XAI canvas should only be seen as an additional aid for the human-centered design process of a self-explanatory AI system. For a successful design process, other methods such as Interviews with users necessary. The XAI Canvas does not provide legal advice, but merely identifies potential legal obligations.
Operating instructions
The idea of the XAI canvas was inspired by the business model canvas and is intended for use in a comparable format. The XAI canvas is used in three steps:
1. Definition of the objectives and benefits of using XAI: list of the overarching objectives to be achieved through the use of XAI and the resulting benefits.
2. Possible legal obligations: Both the General Data Protection Regulation (GDPR) and the planned EU AI Regulation may result in legal obligations that make it necessary to explain AI decisions.
3. Setting up the XAI storyboard: elicitation of user requirements in the form of XAI user stories, which are a modification of the established user stories. Each XAI user story is supplemented with an assessment of the importance of three central characteristics of an explanation (correctness, comprehensibility, time criticality).
The XAI user stories are the main result of the XAI canvas. Based on these XAI user stories, companies can create drafts of the design alternatives in the next step of the design process. The extent to which companies can adequately comply with any legal obligations through the use of XAI must be examined by the companies themselves in a legal consultation.
The XAI guide provides more detailed explanations of the aspects listed in the XAI canvas.
1. Definition of the objectives and benefits of using XAI: list of the overarching objectives to be achieved through the use of XAI and the resulting benefits.
2. Possible legal obligations: Both the General Data Protection Regulation (GDPR) and the planned EU AI Regulation may result in legal obligations that make it necessary to explain AI decisions.
3. Setting up the XAI storyboard: elicitation of user requirements in the form of XAI user stories, which are a modification of the established user stories. Each XAI user story is supplemented with an assessment of the importance of three central characteristics of an explanation (correctness, comprehensibility, time criticality).
The XAI user stories are the main result of the XAI canvas. Based on these XAI user stories, companies can create drafts of the design alternatives in the next step of the design process. The extent to which companies can adequately comply with any legal obligations through the use of XAI must be examined by the companies themselves in a legal consultation.
The XAI guide provides more detailed explanations of the aspects listed in the XAI canvas.
XAI-Canvas - Handout for the human-centered design process
In the field of artificial intelligence (AI), the explainability of AI decisions is becoming increasingly important. With the XAI canvas, KARL provides an overview of relevant aspects of the use of explainable artificial intelligence (XAI). What benefit should be achieved by explaining the AI decisions? To whom should the AI decisions be explained and what requirements do these people have for a helpful explanation?
Target group
The XAI canvas is aimed at all people and companies who want to add explainability components to an AI application in order to better understand AI decisions.
Expected result
The XAI canvas provides support for the initial phases of problem analysis according to the human-centered design process (DIN EN ISO 9241-210): Understanding the context of use and defining user requirements. The XAI canvas supports companies in defining the initial vision for the use of XAI in an AI application. Filling out the XAI canvas first defines the overarching objectives of the declarations and the benefits that users expect from them. The canvas is also intended to raise awareness of any legal obligations that could make the use of XAI necessary. Initial XAI user stories are then created. These provide an initial overview of the different users and their requirements for helpful explanations.
General conditions
It is assumed that an overarching AI use case has already been defined in the company, as this provides the basis for the XAI aspects under consideration. If an AI use case does not yet exist, the KARL AI selection system can help to identify a suitable AI use case for companies. Furthermore, the XAI canvas should only be seen as an additional aid for the human-centered design process of a self-explanatory AI system. For a successful design process, other methods such as Interviews with users necessary. The XAI Canvas does not provide legal advice, but merely identifies potential legal obligations.
Operating instructions
The idea of the XAI canvas was inspired by the business model canvas and is intended for use in a comparable format. The XAI canvas is used in three steps:
1. Definition of the objectives and benefits of using XAI: list of the overarching objectives to be achieved through the use of XAI and the resulting benefits.
2. Possible legal obligations: Both the General Data Protection Regulation (GDPR) and the planned EU AI Regulation may result in legal obligations that make it necessary to explain AI decisions.
3. Setting up the XAI storyboard: elicitation of user requirements in the form of XAI user stories, which are a modification of the established user stories. Each XAI user story is supplemented with an assessment of the importance of three central characteristics of an explanation (correctness, comprehensibility, time criticality).
The XAI user stories are the main result of the XAI canvas. Based on these XAI user stories, companies can create drafts of the design alternatives in the next step of the design process. The extent to which companies can adequately comply with any legal obligations through the use of XAI must be examined by the companies themselves in a legal consultation.
The XAI guide provides more detailed explanations of the aspects listed in the XAI canvas.
1. Definition of the objectives and benefits of using XAI: list of the overarching objectives to be achieved through the use of XAI and the resulting benefits.
2. Possible legal obligations: Both the General Data Protection Regulation (GDPR) and the planned EU AI Regulation may result in legal obligations that make it necessary to explain AI decisions.
3. Setting up the XAI storyboard: elicitation of user requirements in the form of XAI user stories, which are a modification of the established user stories. Each XAI user story is supplemented with an assessment of the importance of three central characteristics of an explanation (correctness, comprehensibility, time criticality).
The XAI user stories are the main result of the XAI canvas. Based on these XAI user stories, companies can create drafts of the design alternatives in the next step of the design process. The extent to which companies can adequately comply with any legal obligations through the use of XAI must be examined by the companies themselves in a legal consultation.
The XAI guide provides more detailed explanations of the aspects listed in the XAI canvas.
Contact person
Robin Weitemeyer | HKA (ILIN)
robin.weitemeyer@h-ka.de
robin.weitemeyer@h-ka.de
Format
PDF
To the offer
