Please enable JavaScript.
Coggle requires JavaScript to display documents.
Data Ethics Committee Guidelines - Coggle Diagram
Data Ethics Committee Guidelines
General ethical and legal principles
Human dignitiy
Self-determination
Privacy
Safety
Democracy
Solidarity and justice
Sustainability
Regulations should not block innovation
Ethical principles don't make regulations obsolete
Perspectives
Data perspective
General requirements
Predictive responsibility
Regard of subject rights
Welfare through the use and sharing of data
Appropriate data quality
Risk-adequate information security
Data quality that suits the use-case
Informational quality, adquate to risk
Interest adequate transparency
Data rights and responsibilities
Power distribution between actors
Rank of individual interest
Contribution of data generation
Rank of individual interests of other actors
Interests of general public
Usage requirements of personal information
Measures against non-ethical usage
Tackle enforcement deficits
Application and specification of the legal framework
Centralization of the
Data protection oversight for the market
Data ownership principle not redommended
Payment methods instead of data as compensation
Personal risk assessment with narrow requirements
Digital estate not enough
Employee data protection
Increase digitalized healthcare
Protection of minors
Data secure design
Methods for data security friendly design
Protection of legal persons and companies
Algorithm perspective
General requirements
Human-centered design
Compatibility with basic social values
Sustainability
Quality and efficiency
Robustness and security
Minimization of bias and discrimination, like machine bias (Angwin et al, 2016)
Transparency, explainability and traceability
Clear accountability structures (also in combination with previous point, see Annany and Crawford, 2018)
System criticality
Damage potential of algorithm
Severity
Probability
Risk adaptive regulatory guidelines
Five criticality stages
Criteria for overarching model
Regulatory measurements
Special measures for high damage potential algorithms
Regular control increased with increase in probability
If damage is unreasonable, the application is forbiddden
DSGVO play central role of governance
Instruments
Follow horizontal
requirements in law (EUVAS), shared with DSGVO
Labeling requirement of system criticality
Proper documentation
Technical and mathematical procedural quality guarantees
Algorithm responsibility (Mittelstadt, 2016)
Algorithm based decisions lead to automatized consequences that may be without human interaction