The last decade has witnessed the rise of a black box society where obscure classification models are adopted by Artificial Intelligence systems (AI). The lack of explanations of how AI systems make decisions is a key ethical issue to their adoption in socially sensitive and safety-critical contexts. Indeed, the problem is not only for lack of transparency but also for possible biases inherited by the AI from prejudices hidden in the training data. Thus, the research in eXplainable AI (XAI) has recently caught much attention. The applications in which AI systems are employed are various. Therefore, there are many requirements for different types of explanations for different users. We survey the existing proposals in the literature by discussing which are the principles of XAI. In addition, we illustrate different types of explanations returned by established explainers. Finally, we discuss their usability and how they can be exploited in real-world applications.

Principles of explainable artificial intelligence

Pedreschi, Dino;Giannotti, Fosca
2021

Abstract

The last decade has witnessed the rise of a black box society where obscure classification models are adopted by Artificial Intelligence systems (AI). The lack of explanations of how AI systems make decisions is a key ethical issue to their adoption in socially sensitive and safety-critical contexts. Indeed, the problem is not only for lack of transparency but also for possible biases inherited by the AI from prejudices hidden in the training data. Thus, the research in eXplainable AI (XAI) has recently caught much attention. The applications in which AI systems are employed are various. Therefore, there are many requirements for different types of explanations for different users. We survey the existing proposals in the literature by discussing which are the principles of XAI. In addition, we illustrate different types of explanations returned by established explainers. Finally, we discuss their usability and how they can be exploited in real-world applications.
2021
Settore INF/01 - Informatica
Explainable AI Within the Digital Transformation and Cyber Physical Systems: XAI Methods and Applications
Springer International Publishing
Ethical data mining; Explainable artificial intelligence; Explanation methods; Interpretable machine learning; Transparent models
File in questo prodotto:
File Dimensione Formato  
Explainable AI Within the Digital Transformation and Cyber Physical Systems_ XAI Methods and Applications _ SpringerLink.pdf

Accesso chiuso

Tipologia: Published version
Licenza: Non pubblico
Dimensione 726.13 kB
Formato Adobe PDF
726.13 kB Adobe PDF   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11384/137139
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 21
  • ???jsp.display-item.citation.isi??? ND
  • OpenAlex ND
social impact