Notes: Solving problems

Design ethnography

  • Method that involves studying people in their natural environment

  • Earliest step in development lifecycle

  • Avoiding silly mistakes and seeing opportunities for innovation

  • Involves:

    • Fieldwork

    • Observing

    • Interviewing

    • Exploring solutions with users (co-design)

    • Etc…

  • Having a diverse number of personas and questioning your assumptions is vital as it can make your product more generalisable to a wider population

  • E.g., A wireless yoga mat: more experienced yoga people may want complex features whilst some inexperienced people may want fewer complex features

Participatory design

  • Using participants (or users) to test your product

  • We can also use personas and scenarios, a real one:

  • E.g., Children using microbit: ideas could lead to any safety or security risk and some included:

    • Kids design reveal potential risks: location tracking and reminders highlighted possible privacy and security risks, such as using a microbit and tying it around a cat’s neck to create a tracker

    • Identifying high-risk data: designs often included sensitive data such as audio, video and location, which helped researchers spot privacy concerns early on

  • Doing this can be great as we can see what users can do and we can quickly highlight any serious concerns involving our products

Risk Mitigation Checklist

Risk Zone 1: Authority and Discipline

  • Will the tech undermine authority?

  • What might the children be able to do?

  • Does it allow a child to escape punishment?

  • Etc…

Risk Zone 2: Malevolence and Accidental Harm

  • How might the tech or data produced by the user be used by a malevolent actor?

  • What does it look like if a user took it too far?

  • What are the risks that a user interacts with a child, knowing or unknowingly?

Risk Zone 3: Emotionality and Socialization

  • Does it appeal to emotionally vulnerable children?

  • Does it isolate them?

  • Does the tech intensify the vulnerabilities?

  • Etc…

Risk Zone 4: Governance and Accounting

  • Will people be producing content that extends the original functionality?

  • What tools would users interact with aa part of normal use of the tech?

  • Is the info presented in a way that promotes informed consent?

Value-sensitive design

  • Designing a system “right” – making sure it respects and supports human values

  • E.g., Automating services: people should be able to make their own choices to reach their goals, so users can have control over relevant features at appropriate times such as changing the volume of something

  • E.g., AI for granting loans: a loan may be granted to someone who is sketchy (false positive) or not giving someone who is trustworthy (false negative), or giving loans to every person or giving none to anyone

Humble AI concept:

  • Promoting fairness and ensuring trustworthy people are not unfairly distributed

Principle of Skepticism

  • Goal = Assume AI is missing info for decision making

  • Approach = Seek additional data to reduce errors

Principle of Curiosity

  • Goal = Learn from borderline cases by testing different decisions

  • Approach = Use “safe exploration” to randomly grant opportunities, e.g., loans, and observe outcomes

  • Outcome = Refines decision threshold over time, making AI more accurate