Group Assignment: Design Concept
Background
In this course you learn about AI, cities, ethics, and (strategic) design. This assignment is aimed at taking the various ideas you are exposed to in the course and applying them in a design project. Only when you can work with these ideas as a designer can you be confident you have understood them to a sufficient degree.
Learning Objectives
After completing this activity, you will be able to create a design concept for a responsible machine learning-based solution for a city-scale issue. More specifically, you will be able to create a system that is contestable, and you will be able to design for the technology of computer vision.
Instructions
- You will be provided with time and space during the Tuesday sessions to work on your design concept with your group.
- During the final session you will be expected to present your design to an external committee.
- Use the workshops in which you are exposed to ML and ethics tools as ways to ground your design in a sufficient understanding of the technology and the ethics involved.
Product
Upon completion of this assignment, you will have created a design concept which satisfies the following criteria:
- Describes a ML-based system that is grounded in a practical understanding of the technology
- Specifically, uses image data collected at street-level
- Is aimed at an urban issue that the municipality of Amsterdam could conceivably be struggling with
- Describes a system that is contestable by design (see the attached poster for a summary of what this means)
The design concept is delivered in the form of:
- Web page: A single page on the course website that briefly describes the concept, and includes the following components:
- Concept video: A short video that visually communicates the front- and backstage human experience of your envisioned system
- Training dataset: The prototype dataset you created to train your model (a ZIP file with images sorted into folders for the various classes)
- Datasheet: A PDF document that describes critical aspects of your dataset
- Model: The prototype model that you trained using the aforementioned dataset (a link to the model uploaded to the google cloud from teachable machine suffices)
- Model card: A PDF document that describes critical aspects of your model
- Final presentation: A slide deck in PDF format
- Annotated portfolio: A booklet containing the various artifacts produced throughout the design project, with annotations indicating what each shows
Follow-up
- The project web page is published publicly only under a Creative Commons Attribution licence
- You will use the slide deck during the final session to deliver your presentation; you will receive informal feedback from the audience immediately after the presentation
- You will deliver the annotated portfolio at the end of the final week, you will receive written feedback from your instructors some time after delivery
- You will use all the material you produce throughout this assignment as the foundation for your individual paper
- The design project itself does not count towards your final grade