In this module, we’ll consider how the complexity of distributed, smart, and connected systems challenges explainability, but also provides a resource for critical design. In this exercise you’ll take cues from the work of Graham Dove & Anne-Laure Fayard, and explore the how the supernatural can be useful metaphor to examine how people understand and intrepret everyday systems. The authors note:
“Research into ML often focuses on explaining algorithmic decision-making or making it more transparent to users. However, transparency and explanations have limitations … Monsters are to be feared, but also are generative spaces, places to question, wrestle with uncertainty, resist easy classifications, name power.”
In this study, “Monsters, Metaphors, and Machine Learning”, the authors developed a hands-on workshop to materialize questions posed by machine learning. They introduced a series of ‘monster cards’ (see above) to help encourage reflection on designing with machine learning. Each monster is related through a brief description to an aspect of machine learning’s uncertainty and processes. Using this, participants map out their assumptions and concerns about these processes and later materialize them as a monster using everyday materials.
As part of this exercise, you’ll explore questions of uncertaintly around systems, first-hand, and the value of ‘spookiness’ as a metaphorical approach by adapting/repeating the exercise yourself.
Brief: Drawing on the examples introduced in class, work with someone to examine an everyday technology of the smart and connected home. Use the supernatural or superstitious metaphor to help this exploration. Prepare a materialization or diagram their mental models.
Note This is a warmup exercise and you should spend around 2 hours on this exercise.
Note We’re still in COVID land. This project is to be completed remotely / safely and social distanced!
As part of this exercise you will be asked to:
You are asked to deliver three things for this warm up exercise:
The narrative and reflection should be approx 150-200 words (max.)
Share your outcomes as a post on the #projects channel of on Slack.