Our workshop envisions what we call “robotic home”: how smart home technology could be used as a tool for increasing mental health. We will introduce the robotic home approach that is proactive and interactive with the inhabitant. We focus on robotic homes as enhancing mental health both through therapeutic means and by assisting in the aspects of regular life that the inhabitant may have difficulties with because of their condition. The very home may effectively become a behavioral therapist. We explore what kind of ethical risks are involved, and how to minimize the risks. 

This talk introduces the idea of a robotic home and discusses some ethical aspects of the design for it. “Robotic home” is a version of smart home where the AI systems and the physical devices they control are active towards the inhabitant(s). While a smart home has AI mainly as a tool for convenience, a robotic home is a cluster of both embodied (robotic) and disembodied AI systems that interact with the inhabitant(s) of the home. The context for the discussion is an environment that is designed as a whole to assist a person who requires active assistance for everyday life (for instance, for 
mental health reasons, or disability) and the home itself takes over some functions and active control needed for this. A living environment such as this should be a coherent, functional whole where the different parts of the design serve the special needs of the inhabitant, as well as moral values such as dignity and autonomy. A version of Aimee Van Wynsberghe’s (2015) care centered value-sensitive design model of care robotics is presented as a practical ethical framework for designing this environment, including the analysis of personalized needs for assistance, (physical and non-physical) tasks that the robotic home is responsible of, and ethical considerations. 

Ambient technology such as AI assistants demonstrate that when equipped with motor and sensory capabilities, AI can meaningfully contribute to quotidian activities, such as staying on top of household tasks and appointments, eating regularly, and maintaining sleep hygiene. Given that executive dysfunction, which generates difficulties in daily activities, is present in most common psychiatric diagnoses (including anxiety disorders, mood disorders, and ADHD), it is unsurprising that psychiatric subjects often struggle to stay on top of daily tasks. These struggles in turn negatively impact well-being. In this paper, I discuss the prospects of situated AI in helping steer daily activities, in effect acting as a behavioral therapist. To be successful, this requires responsive technology that predicts needs rather than responds only to user-generated reminders. The development of such algorithms, however, requires substantial ethical oversight. 

In my talk, I suggest that when we are talking about robotic and AI technologies such as smart home devices, an extended view of cognition is helpful. The theories of extended cognition (see e.g. Clark 2008) and AI-extenders (Hernandez-Orallo & Vold 2019) are useful theoretical tools for investigating extended or hybrid systems, such as a person using smart home technology as a part of their coping with activities of daily living. The role of such external resources becomes acute with people in vulnerable situations, especially with people who suffer from psychiatric disorders, or elderly people. When one adopts the “extended” view, the integrated technologies are counted as part of the realising basis of their users cognitive and affective capacities. This entails that the robotic home resources a person is reliant on should be given a stronger 
ethical and legal status, and the patient should be allowed to have access to it e.g. while she/he is being diagnosed. I suggest that especially for ethical reasons, we should treat certain robotic equipment as part of their users’ cognition and mind. 

I argue that responsible implementation of social robots in mental healthcare requires that they are both designed to be culturally competent and their socially pernicious effects are constrained. I assert that this requires an ameliorative approach to the mental health concepts and values embedded in automated healthcare robots. First, responsible design should aim to make the embedded concepts and values in the robots culture-sensitive, while simultaneously aim to avoid reinforcing commonly held biases and harmful values, such as racism and sexism. Second, the design and implementation of healthcare robots should also be evaluated based on their indirect social consequences that go beyond their predesigned application. These consequences can be understood in light of what Ian Hacking (1999) has called the looping effects of interactive kinds, when medical practices influence the labeled people to such a degree that the practices have to be amended. In particular, I argue that the benefits of implementing mental healthcare robots in different cultural contexts should be evaluated based on whether they can avoid distorting local conceptions and inflicting stigma. Finally, I apply my analysis to argue for an ameliorative approach to designing and implementing “robotic homes” so as not to inflict unintended social and health effects.