SIENNA response to the French Committee for Digital Ethics public consultation on the ethics of conversational agents

2020-11-12

On 30 October 2020, the SIENNA project submitted its response to the French National Pilot Committee for Digital Ethics (CNPEN) public consultation on the ethics of conversational agents, what is more commonly known as 'chatbots'. Our key recommendations were to reduce the anthropomorphisation of chatbots and conduct impact assessments to identify risks and harms. 

Person chatting on a mobile phoneThe response was structured around twenty issues, ranging from questions on trust in chatbots, chatbot surveillance, emotion detection, and chatbot use in workplaces. Our key recommendation was to reduce as much as possible the confusion of a chatbot with a human being, thus minimising the potential for anthropomorphisation and emotional attachment to a machine. Specific recommendations included always informing a human when they are interacting with a chatbot, using non-human identifiers in the design of chatbots, and always providing humans with the opportunity to interact with another human when requested. Furthermore, to address the risk of reproducing and reinforcing gender stereotypes, especially female subservience, we recommended the use of androgynous identities and voices.

Throughout our feedback, we encouraged more social sciences studies and impact assessment on human-chatbot interaction to understand and address the potential and actual impacts more concretely. Furthermore, we called for sufficient testing of chatbots before deployment to identify risks and harms. This is especially true when use of a chatbot might impact vulnerable groups (e.g. elderly or children), as governments have legal obligations to protect vulnerable people including under human rights law. There are contexts where red lines should be established; one such potential red line is the use of chatbots in AI-enabled toys for children.

As with the governance of any emerging technology, we urged that chatbots must be regulated through law that protects against harms and guarantees human rights, including access to complaint and redress mechanisms.

More about the French National Pilot Committee for Digital Ethics (CNPEN) public consultation on the ethics of conversational agents on their website. 

Read SIENNA's response

By Nicole Santiago, Trilateral Research

News from SIENNA