July 26, 2021
By Ahmed Bouzid (Founder & CEO at Witlingo)
1. The voicebot is a tool
The voicebot is first and foremost a piece of technology and it exists to make the life of a human being better. The voicebot is not a human being and it should never pretend to be a human being or claim the privileges of a human being. It can mimic a human being, and it can lean on the techniques that human beings use with each other to communicate effectively. But it should not behave in such a way that the user is fooled into thinking that the voicebot is as sophisticated and as emotionally intelligent as a human being. Setting aside the ethical issues that such a voicebot would introduce (for instance, using politeness protocols to gain the trust of the user to the point that the user forgets that the voicebot is merely a tool that a private corporation or a government agency is using to enable the user to buy things or provide sensitive information), behaving in a way that tricks the user into thinking that the voicebot is much smarter than it really is will lead to the user conversing in a way that outsrips the ability of the voicebot, leading to frustration and disappointment.
2. The voicebot should be as smart as the data that it has
The voicebot should make the most of the information it has about the user and the context of its interaction with that user: What are the user’s preferences; what are the preferences of most users who are engaging with the voicebot; what patterns of behavior is it aware of that it can leverage to anticipate and interpret the meaning of what users say and do; what are the physical constraints within which the user is interacting with the voicebot; what are the social constraints (is anyone else in the room while the voicebot and the user are talking) or the psychological constraints that are in effect?
3. The voicebot should be consistent
The voicebot should not use different words for the same entity in its language with the human. It should also stick with only one voice and change that voice only when it makes sense (for instance, the voicebot is telling a story with multiple characters). It should not change the language register that it uses (for instance, going from formal to informal).
4. The voicebot should be transparent
The human should always know if they are speaking with another human, with a voicebot, or with a voicebot assisted by humans behind the scene. Again, setting aside ethical issues, a user who knows that they are speaking with a voicebot will probably behave in a way that will increase the likelihood of a successful interaction than one who thinks that they are engaged with a human being.
5. The voicebot should be respectful
The voicebot should never be disrespectful to the user. What does it mean for a voicebot to be respectful to a user? Observing the Cooperative principle and the five Conversational Maxims would be a good first start.