W2 Lec Notes

Lec 1

question about whether technological artifacts have morals imbedded in them or not:

• is technology neutral or malicious/good
  • same thing like are guns evil

Theres a lot of power in those who design technology:

• the gadget you are creating → does the responsibility rely on me as the creator or the user.

Kranzberg’s laws:

• technology is neither good nor bad nor neutral:
  • tech has social, environmental and human consequences that go far beyond the immediate purposes of the technical devices.

Values, principles, purpose, responsibilities, practices:

• they are all different
• Values - thing that we think are important/strive for
• principles - things if ou believe are high value and do them in practice
• translates into responsibilityies to achieve valuex/expectations
• Practices - similars

Ethical principles typically get put in place way before the law:

• can even lead to new laws

Code of ethics → set of rules about good and bad behaviour To be a licensed profession, you have to abide by a certain set of ethics

Can an action be ethical but illegal or legal but unethical

You need to consider different stakeholders:

• you will impact more ppl than just the target audiencce of products you design.

Beneficence vs Non-Maleficence:

• Do good vs Do no harm
• Do good AND do no harm

Talking about deploying certiain things:

Argument for neutrality: “This moral neutrality is based upon viewing technology purely as a means (providing tools for society to use) with the ends (the actual usage of technology) lying beyond and outside the realm of engineering;

**Automated deicsion making:

• A computerised process that assists/replaces judgement of human decision makers

L2:

Therac 25 Value neutrality thesis:

• whether technology has value assigned to it Why would u trust someone: honest, repuation, loyalty, etc…

FUll automation with full control Tension between automation and human control

• make sure that we can still control the automation → inc case there is an error in the automation

Accountability - 4 barriers:

• problem of many hands
  • when u design software, deisgn by many ppl
• problem of bugs
• blaming the computer
• software ownership without liability

ACM principlaes for alogirthmic transparency and accountability

1. Awareness
2. Access and redress
3. Accountability
4. Explanation
5. Data provenance
6. Auditability
7. Validation and testing

We want high computer automation and high control → reliable safe and turstowrhty

Where does AI ethics sit? Factors:

• human social and envrionmental weelbeing
• reliability
• human -centred values
• transparency and explainability
• fairness
• contestability
• privacy protection and security
• accountbility

AI alignment - design ai to align with the norms and values of your user group in mind

Personas

• models of users
• Rpresent a group of typical users
• can’t be based on a real person
  • but can be a ccharacicature that represents the typical person from this model of users

Constructing:

1. Identify Behavioural variables
2. map interview subject to behavioural variables
3. identify significant behaviour patterns
4. synthesize characteristics and relevant goals
5. check for completeness and redundancy
6. Designate persona type

Value sensitive design: an approach that rigorously accountd for human values in the technical design and engineering process

VSD

• Identify the stakeholders
• surfacing of their values through conceptual and empirical investigation

The Judgement Call - The Game

conceptual investigations https://youtu.be/LXqlAXEMGI0

• we should do this in the assignment
• Talk about a scenario
  • Ethical Principals
    • fairness
    • privacy and security
    • reliability and safety
    • transpareny
    • inclusion
    • accountability
  • Rating

Stakeholders:

• direct - use the technology or being used on them
• indrirect - feel the impact
• sepcial populations - ppl who can’t use t or choose not to

Presnetation of the product:

• who will use it
• what features will it have?
• when will it be used?
• where will it be used?
• why is it useful?
• how will it be used?

Questions to ask when consdidering things:

• who is the most impacted
• what features are the most problematic
• what are teh potenntial harms

Things to consider:

• location data
  • e.g pokemon go making spawn places more common in certain areas for certain demogrphics
• displayed info
  • facebook manipulating peoples mood by filtering their feeds
  • even tho ppl consented to their data being collected, not consented to be experimented on though
  • users should consent - e.g be told that it will happen to avoid the observer effectr