Data Governance for Ethical AI

There are three types of lies: lies, damned lies and statistics

??

There are three types of lies: lies, damned lies and statistics

Benjamin Disraeli

There are three types of lies: lies, damned lies and statistics

Benjamin Disraeli 1804-1881

There are three types of lies: lies, damned lies and ‘big data’

Neil Lawrence 1972-?

Mathematical Statistics

‘Mathematical Data Science’

Embodiment Factors

Evolved Relationship with Information

New Flow of Information

Evolved Relationship

Evolved Relationship

The Gartner Hype Cycle

Cycle for ML Terms

Google Trends

❮ ❯

Challenges

1. Paradoxes of the Data Society

2. Quantifying the Value of Data

3. Privacy, loss of control, marginalization

Challenges

1. Paradoxes of the Data Society

2. Quantifying the Value of Data

3. Privacy, loss of control, marginalization

Privacy, Loss of Control and Marginalization

• Society is becoming harder to monitor
• Individual is becoming easier to monitor

Digital Revolution and Inequality

• Potential for explicit and implicit discrimination on the basis of race, religion, sexuality, health status
• All prohibited under European law, but can pass unawares, or be implicit
• GDPR: General Data Protection Regulation

Discrimination

• Potential for explicit and implicit discrimination on the basis of race, religion, sexuality, health status
• All prohibited under European law, but can pass unawares, or be implicit
• GDPR: Good Data Practice Rules

Marginalization

• Credit scoring, insurance, medical treatment
• What if certain sectors of society are under-represented in our analysis?
• What if Silicon Valley develops everything for us?

Digital Revolution and Inequality?

Data Science Africa is a bottom up initiative for capacity building in data science, machine learning and AI on the African continent

Example: Prediction of Malaria Incidence in Uganda

• Work with Ricardo Andrade Pacheco, John Quinn and Martin Mubaganzi (Makerere University, Uganda)
• See AI-DEV Group.
• See UN Global Pulse Disease Outbreaks Site

Malaria Prediction in Uganda

Tororo District

Malaria Prediction in Nagongera (Sentinel Site)

Mubende District

Malaria Prediction in Uganda

GP School at Makerere

Kabarole District

Early Warning System

Early Warning Systems

Feudal Era Data Ecosystem

Information Barons threaten our Privacy

African Data Sharing Covid-19

Amelioration

• Work to ensure individual retains control of their own data
• We accept privacy in our real lives, need to accept it in our digital
• Control of persona and ability to project
• Need better technological solutions: trust and algorithms.

Data Governance Toolkit

Society

Vulnerabilities

Enfranchisement

Individuals

Personal Data Trusts

Motivation

1. Indsidious decision-making that has downstream instrumental effects we don’t control.
2. A power-asymmetry between data-controllers and data-subjects
3. A loss of personhood in the re-representation of ourselves in the digital world.
4. The GDPR’s endeavour to curb contractual freedom cannot by itself reverse the power-asymmetry between data-controllers and data-subjects.

Analogy

• Digital Democracy vs Digital Oligarchy Lawrence (2015a) or Information Feudalism Lawrence (2015b)
• Data subjects, data controllers and data processors.

Legal Mechanism of Trusts

• Fiduciary responsibility of Trustees.
  • Burden of proof in negligence is reversed.
• Trustees are data controllers
• Beneficiaries are data subjects
• Power of data accumulation wielded on the beneficiaries behalf
• See Edwards (2004), Delacroix and Lawrence (2019) and Lawrence (2016)

Data Trusts Initiative