PhD Anticipating the Dynamics of Learning Systems in the Home Context
PhD Anticipating the Dynamics of Learning Systems in the Home Context
You cannot apply for this job anymore (deadline was 15 Jun 2020).
Browse the current job offers or choose an item in the top navigation above.
Prospective starting date: August 2020
Job description
Background
The Future Everyday Group of the Industrial Design Department of TU/e invites applications for a fully funded 4-year PhD project. The aim of the project is to explore the role of learning systems in the dynamics of everyday life using a social practice theory/co-performance perspective and a research through design approach.
Equipped with sensors, actuators and processing power, smart home systems make context aware decisions. For example, thermostats measure the room temperatures to decide whether to switch the heating on. A problem with these systems is that they have trouble adjusting their actions to the specific circumstances. This can lead to mistakes. For example, heating a room while nobody is there. In an attempt to better cater for unique circumstances and preferences, smart home systems increasingly include some form of learning over time. Learning thermostats, for example, autonomously adjust their heating pattern to a household by learning their weekly schedule and preferences. However, this is often not turning out as intended. Among a range of other reasons, learning systems have trouble in distinguishing between regular and exceptional events, resulting in erroneous patterns that are difficult to correct by the users. To address these problems, ongoing research efforts are trying to make algorithms understandable, provide better and more contextualised information, and offer users more control over the learning of their devices.
Challenges
While this is a positive development, the work so far misses a dynamic perspective on everyday interactions between users and these learning systems at home. Over time, the roles of both systems and users are expected to change and reshape everyday life in new ways. For example, studies show that people with learning thermostats sometimes adjust their schedules to the erroneous patterns of their devices. A focus on these longer-term processes gives rise to new questions. For example: How does everyday life change as both smart home systems and users learn over time? And what are the roles of the designers of the algorithm (and its updates) in this dynamic in different phases of usage? Is there a risk of everyday life reshaping to fit artificial preferences for predictability and measurability? What will, and could, everyday life be like when learning algorithms become increasingly integrated into it?
Approach
Social practice theories, as developed by Schatzki, Shove and colleagues, and in the related concept of co-performance developed by Kuijer form useful theoretical perspectives to identify and understand these processes. Using the unit of practices (cooking, bathing, sleeping, parenting, etc.), the theory looks beyond product-user interactions, while staying at a level that is meaningful for designers. Moreover, by centralizing the recursive relation between everyday actions and collective ideas of normal conduct, it helps to identify and understand larger mechanisms of social change in terms of concrete human and artificial actions. The related concept of co-performance, which places smart technologies next to humans as co-performers of practices, helps bring the designer of the algorithm into everyday life dynamics.
Research through design is highly suitable to address these types of complex, future-oriented questions. Through making and creative imagination, design enables anticipation of and debate around possible futures. The project is envisioned to contain three main phases. The first focuses on the probable dynamic effects of further integration of learning systems into the home. The second phase focuses on opening possible roles of learning systems in everyday life through a range of experimental designs, and their deployment in everyday life. This phase makes use of the smart home test environment developed at Industrial Design Eindhoven and will make experimental use of platforms like Home Assistant and OpenHAB. The third phase focuses on preferred futures for learning systems in everyday life. It builds on work in human-in-the-loop and explainable AI to envision democratic, inclusive and sustainable futures of human-computer collaboration.
At the end of the 4-year project, the candidate is expected to defend their PhD Thesis. During the project the candidate is expected to publish in relevant scientific conferences and journals within the field of human-computer interaction, such as the CHI and DIS conferences, Ubicomp and the ToCHI journal. Moreover, the candidate is supported and stimulated to engage with the practice theory research community through the taking of courses, research exchanges and conference presentations at 4S/EASST, RGS and Anticipation.
There is an option to extend the PhD position into a 5-year position with additional teaching responsibilities and training.
The team
The PhD will be supervised by Dr. Lenneke Kuijer and Dr. Mathias Funk in the Department of Industrial Design at Eindhoven University of Technology, a world leading department in human-computer interaction and widely known for its design research. Lenneke Kuijer is an expert in the areas of social practice theory, co-performance and research through design. The PhD position is connected to her VENI project 'Anticipating the role of smart technologies in the dynamics of everyday life', a prestigious fellowship funded by the Dutch Organisation for Scientific Research. Mathias Funk has a background in computer science and researches design tools and methods for data-intensive products and systems of smart things. Both are members of the Future Everyday Group. The PhD position will be associated with the Eindhoven AI Systems Institute (EAISI) and the Human-AI alliance with Utrecht University.
About Eindhoven University of Technology
Eindhoven University of Technology (TU/e, www.tue.nl) is one of Europe's leading research universities. The Eindhoven area, in the southern part of the Netherlands, is one of Europe's top 'innovation ecosystems', with many high-tech companies and institutes. TU/e is intertwined with many of these companies and institutes, and research at TU/e is characterized by a combination of academic excellence, industrial relevance and societal interweaving. The Department of Industrial Design (ID) of the Eindhoven University of Technology (TU/e), founded in 2001, is a maturing department with over 650 students, both Bachelor and Master, and around 40 research staff members and about 10 lecturers. The mission of the department of Industrial Design at TU/e is Research on and Education in the Design of Systems with Emerging Technologies in a Societal Context.
Prospective starting date: August 2020
The Future Everyday Group of the Industrial Design Department of TU/e invites applications for a fully funded 4-year PhD project. The aim of the project is to explore the role of learning systems in the dynamics of everyday life using a social practice theory/co-performance perspective and a research through design approach.
Equipped with sensors, actuators and processing power, smart home systems make context aware decisions. For example, thermostats measure the room temperatures to decide whether to switch the heating on. A problem with these systems is that they have trouble adjusting their actions to the specific circumstances. This can lead to mistakes. For example, heating a room while nobody is there. In an attempt to better cater for unique circumstances and preferences, smart home systems increasingly include some form of learning over time. Learning thermostats, for example, autonomously adjust their heating pattern to a household by learning their weekly schedule and preferences. However, this is often not turning out as intended. Among a range of other reasons, learning systems have trouble in distinguishing between regular and exceptional events, resulting in erroneous patterns that are difficult to correct by the users. To address these problems, ongoing research efforts are trying to make algorithms understandable, provide better and more contextualised information, and offer users more control over the learning of their devices.
Challenges
While this is a positive development, the work so far misses a dynamic perspective on everyday interactions between users and these learning systems at home. Over time, the roles of both systems and users are expected to change and reshape everyday life in new ways. For example, studies show that people with learning thermostats sometimes adjust their schedules to the erroneous patterns of their devices. A focus on these longer-term processes gives rise to new questions. For example: How does everyday life change as both smart home systems and users learn over time? And what are the roles of the designers of the algorithm (and its updates) in this dynamic in different phases of usage? Is there a risk of everyday life reshaping to fit artificial preferences for predictability and measurability? What will, and could, everyday life be like when learning algorithms become increasingly integrated into it?
Approach
Social practice theories, as developed by Schatzki, Shove and colleagues, and in the related concept of co-performance developed by Kuijer form useful theoretical perspectives to identify and understand these processes. Using the unit of practices (cooking, bathing, sleeping, parenting, etc.), the theory looks beyond product-user interactions, while staying at a level that is meaningful for designers. Moreover, by centralizing the recursive relation between everyday actions and collective ideas of normal conduct, it helps to identify and understand larger mechanisms of social change in terms of concrete human and artificial actions. The related concept of co-performance, which places smart technologies next to humans as co-performers of practices, helps bring the designer of the algorithm into everyday life dynamics.
Research through design is highly suitable to address these types of complex, future-oriented questions. Through making and creative imagination, design enables anticipation of and debate around possible futures. The project is envisioned to contain three main phases. The first focuses on the probable dynamic effects of further integration of learning systems into the home. The second phase focuses on opening possible roles of learning systems in everyday life through a range of experimental designs, and their deployment in everyday life. This phase makes use of the smart home test environment developed at Industrial Design Eindhoven and will make experimental use of platforms like Home Assistant and OpenHAB. The third phase focuses on preferred futures for learning systems in everyday life. It builds on work in human-in-the-loop and explainable AI to envision democratic, inclusive and sustainable futures of human-computer collaboration.
At the end of the 4-year project, the candidate is expected to defend their PhD Thesis. During the project the candidate is expected to publish in relevant scientific conferences and journals within the field of human-computer interaction, such as the CHI and DIS conferences, Ubicomp and the ToCHI journal. Moreover, the candidate is supported and stimulated to engage with the practice theory research community through the taking of courses, research exchanges and conference presentations at 4S/EASST, RGS and Anticipation.
There is an option to extend the PhD position into a 5-year position with additional teaching responsibilities and training.
The team
The PhD will be supervised by Dr. Lenneke Kuijer and Dr. Mathias Funk in the Department of Industrial Design at Eindhoven University of Technology, a world leading department in human-computer interaction and widely known for its design research. Lenneke Kuijer is an expert in the areas of social practice theory, co-performance and research through design. The PhD position is connected to her VENI project 'Anticipating the role of smart technologies in the dynamics of everyday life', a prestigious fellowship funded by the Dutch Organisation for Scientific Research. Mathias Funk has a background in computer science and researches design tools and methods for data-intensive products and systems of smart things. Both are members of the Future Everyday Group. The PhD position will be associated with the Eindhoven AI Systems Institute (EAISI) and the Human-AI alliance with Utrecht University.
About Eindhoven University of Technology
Eindhoven University of Technology (TU/e, www.tue.nl) is one of Europe's leading research universities. The Eindhoven area, in the southern part of the Netherlands, is one of Europe's top 'innovation ecosystems', with many high-tech companies and institutes. TU/e is intertwined with many of these companies and institutes, and research at TU/e is characterized by a combination of academic excellence, industrial relevance and societal interweaving. The Department of Industrial Design (ID) of the Eindhoven University of Technology (TU/e), founded in 2001, is a maturing department with over 650 students, both Bachelor and Master, and around 40 research staff members and about 10 lecturers. The mission of the department of Industrial Design at TU/e is Research on and Education in the Design of Systems with Emerging Technologies in a Societal Context.
Prospective starting date: August 2020
Specifications
- max. 38 hours per week
- Eindhoven View on Google Maps
:fill(white)/logos/tue-en-wide_acOLN2b.png)
Eindhoven University of Technology (TU/e)
Requirements
- An MSc degree in Interaction Design or closely related discipline
- Familiarity with contemporary machine-learning approaches and a drive to understand their application in detail, basic Python or Java programming skills are required, full computer science degree or related expertise are not expected
- Proven keen interest in social theory and at least basic familiarity with social practice theories
- A critical attitude towards the role of technology in everyday life
- Proven skills in design research, in particular research through design/constructive design research and experience with academic paper writing
- Good collaboration and communication skills, both with other disciplines and outside academia
- Fluency in spoken and written English language
- A willingness for international travel and (extended) stays abroad
Conditions of employment
- We offer a full time, challenging position for 4 years (48 months) in a new and rapidly growing Department of Industrial Design.
- We offer a starting salary of € 2,325 gross per month in the first year and € 2,972 gross per month in the last year (PhD-salary scale Collective Labour Agreement Dutch Universities) on a full-time basis, plus 8% holiday allowance and 8,3% end of the year allowance.
- Attractive secondary labour conditions (good sport facilities, child-care, etc.).
