The aim of this paper is to contribute to the debate on ethics in AI focusing particularly on the social and ethical issues which emerge during human–conversational agent interaction. Those issues will be approached adopting both the perspective of the user and the one of the machine. In fact, it is often stressed the need to include the ethical component within system's design, which it is certainly necessary as will be discussed in the article. Nevertheless, conversational agents are tools, although they are capable of having a strong impact on the society; thus, the moral use of these social tools must be demanded to users as well. For this reason, it will be proposed a new set of ethical guidelines, called Mirror Ethics, based on the alignment of both parties involved in the conversation to a shared moral baseline. The adoption of these shared values will lead to the creation of a new ethical equilibrium between users and conversational agents, producing a social beneficial impact and increasing the level of trust within the human–machine interaction.

本文的目的是促进有关AI伦理学的辩论，特别是关注人与人之间的互动过程中出现的社会和伦理问题。这些问题将通过用户和机器的角度来解决。实际上，通常会强调需要在系统的设计中包括道德成分，这在本文中将会讨论，这当然是必要的。尽管如此，对话主体还是工具，尽管它们能够对社会产生深远的影响。因此，也必须要求使用者道德使用这些社交工具。出于这个原因，将基于对话双方共同遵循共同的道德底线，提出一套新的道德准则，称为“镜像伦理”。