Preface

• Pages: 11-12

11

PREFACE



the goal has been for machines to have human behaviour. This has been

translated into technical jargon that uses illusory terms such as learning in-

stead of tuning, training instead of iterative loop, and intelligence instead of

advanced pattern extraction. In general these terms are concise and facilitate

communication but convey an underlying anthropomorphic idea that con-

fuses non-experts and sometimes even experts.



of application in which amazing simulations of human performance have

been achieved, have raised concerns about how AI systems should perform

when personal data is used or their results impact the lives of human beings.

Various problems have arisen when applying these systems to tasks that re-

quire human judgement. The most common response has been the develop-

ment of ethical frameworks and guidelines, which essentially consist of more

or less detailed lists of ethical principles, such as European Union’s Ethics

Guidelines for Trustworthy AI (2019) and UNESCO’s Recommendation on



Ethics is currently understood as the set of moral rules that govern a per-

son’s conduct in any sphere of life. If AI is intended to replace or at least

simulate human judgement, it makes sense to speak of AI ethics. This book

examines whether the ethics of AI embodied in such ethical principles are



also takes the same approach.

The European Comission’s High-Level Expert Group for AI, after studying

Ethics

Guidelines for Trustworthy AI. Compliance with a set of principles, compli-

ance with applicable laws and sound reliability are the conditions set out in

these Ethical Guidelines for a system to be considered trustworthy. The term

trustworthy adds confusion to the conversation, since in common parlance it

has different meanings depending on where it is used.

  

Tech’s drive to create new AI-driven systems are driving the adoption of dif-

ferent ways of solving problems that are generating controversy. The trust-

worthiness rating of these models, when deployed in high-risk environments,