From ethical AI frameworks to tools: a review of approaches AI and Ethics

In addition to user consent, companies also display links to their privacy policies, enabling users to access detailed information about how their data is collected, used, and managed. These privacy policies outline the data handling practices and user rights in line with applicable laws and regulations. Sentiment analysis, the measurement of the positiveness or negativeness of the user messages, can be an indicator of customer satisfaction.

The various tools differ substantially regarding the point of intervention in the design process (c.f. Fig. 1). This is an important aspect for understanding how they relate to the AI development process.Footnote 3 Frameworks, algorithms, software libraries, etc., usually aim at the ex-ante creation of an ethical system. Audits, checklists, and metrics typically are instruments applied to an AI system once it has been developed, potentially to improve its ethicality in an iterative fashion. Information about a system, declarations and labels can be applied ex-post, including in cases where a comprehensive ethical system is not possible. Communities and data sets can be considered infrastructure supporting the various stages of development. Although true conversational AI chatbots are not rule-based or scripted, most still do rely upon some scripted answers for specific queries.

Key components of responsible AI implementation:

Companies should make their AI systems accessible to all users, including those with disabilities or specific needs. This can be achieved by providing different interaction alternatives, such as voice input, text input, or touch input. Companies should also consider providing content in various formats, such as audio, transcripts, or visual cues, to cater to different user preferences and accessibility requirements.

Responsible AI is guided by several key principles that organizations adopt to ensure ethical and trustworthy development and deployment of AI systems. These principles promote transparency, fairness, privacy, security, reliability, and sustainability in AI applications. By adhering to these principles, organizations can respect human values and rights while building AI systems that benefit individuals and society as a whole. Note that some approaches are relevant in a certain development step, but do not necessarily support that step in the development process.

Analyzing historical data for Intents

Other topics and more general aspects are often addressed with conceptual frameworks and process models. All references were analyzed with respect to the approach and the ethical aspects that it addresses. Many papers address more than one ethical issue (e.g. privacy and explainability) and some propose more than one approach (e.g. an algorithm and software). Based on the resulting list, the categorization was refined to eliminate categories with only a few or no references and to better group the categories as presented below.

Read more about What Are the Ethical Practices of Conversational AI? here.