Kamili Afra Reference Sheet Dec 11, 2024
Data Gov. for Dummies (Jonatahan Reichental)
●It’s all about managing data well (increase data
value and reduce risk)
●Leadership and Strategy: align w/ vision
organization
●Roles and resp: find the right team member
Data gov a guide (Dimitrios Sargiotis)
●Guideline: set of recommendation that provide
high-level guidance
●Policies: formal documents that outline rules,
proecdures, and requirements Edquist et al (Sept 2022)
●Standards: detailed specifications and technical the redistribution of responsibilities, the reality gap,
requirements opaque trade-offs between privacy and fidelity, the
●Practices: actual methods, processes, and actions reappraisal of transparency and the myth of reduced
to implement data gov. bias are just a few examples of the challenges we
●Procedures: step-by-step instructions must consider when it comes to the future use(s) of
●Data governance framework: Governance SD (Synthetic data).
Structure, Data Policies and Standards,
Decision-Making Processes, Roles and Ashleigh McCabe
Responsibilities, (Monitoring, enforcement, and ● Big data ethics: informed consent, privacy,
continuous improvement) ownership, algorithm bias and objectivity, big
data divide (current state of data access)
Data Ethics and Challenges (Samiksha Shukla, Jossy Shanley, D. et al (2024)Implementing data ethics is
P. George, Kapil Tiwari, Joseph Varghese Kureethara) challenging but essential, as failure risks losing
Data gov. customer trust and long-term organizational
●Types of data: nominal, ordinal, ratio, internal value.●Don’t forget about the sources●ROI vs
●Big data: value, volume, variability (undergoing Impact●Data Scientist is not an Ethicist●Set Rules,
continuous change), visualization, veracity Communicate AND enforce values, Incorporate
(accuracy), variety, velocity Diversity, Identify Champions, Understand impact
●FAIR principle: findable, accessible, interoperable,
reusable Ethics:
Data Ethics ●Ethics (group) vs morals (personal)
●To Consider:●What happens when personal ethics ●Branches of Ethics: ●Applied Ethics - Applies
conflicts with professional ethics?●Who owns what ethical theories to real-life issues.○Example:
data? Consider the supply chain●How do firms Should euthanasia be legalized?●Meta-Ethics -
provide informed consent?●Who and how to Examines the nature and meaning of ethical terms
define common and professional ethics regarding and judgments.○Example: What does "good" truly
data? mean?●Normative Ethics - Studies standards for
right and wrong behavior.○Example: Is lying ever
Data Privacy morally acceptable?●Debate if its own branch or
●how data is shared, stored, and used across part of above○Descriptive Ethics - Observes and
different channel describes people's moral beliefs without
●What for: legal responsibility, earn employee trust, judgment.■Example: Surveying attitudes toward
prevent hacking, secure assets animal rights.
●Principle: Notice or Awareness; choice or concent; ●Ethical theory:
access or participation; integrity or security; ○Consequentialism: a theory of normative ethics
enforcement or redness that states that the moral value of an action or
●Threats of Data Privacy: Behaviour Tracking and decision should be judged based on its
Liberalization of Data Market, Misuse of Data consequences.
Transparency ○Deontology: ethical theory that suggests an
individual should act based on what they
believe to be morally right, regardless of the
consequences
Kamili Afra Reference Sheet Dec 11, 2024
○Virtue Ethics: moral philosophy that ●Difference between standards, frameworks, laws,
emphasizes individual’s character and and regulations
personality traits instead of their actions ○Purpose
●Data science ethics: both the existence of a choice ○Voluntary vs Mandatory
and the moral implications of the practitioner’s ○Scope and Applicability
ultimate decision are key aspects of each data ○Flexibility
science;; both the existence of a choice and the ○Development and Governance
moral implications of the practitioner’s ultimate ○Enforcement and Compliance
decision are key aspects of each data science
●Data ethics 5C: Consent, collection, control, Algorithmic and Cognitive Biases
confidentiality, compliance ●Data bias: unrepresentative training data
●Importance of Data ethics: Protects Public Trust, ●Leavy et.al: “the implementability of data
Promotes Transparency, Prevents Bias, governance regulations to ensure bias-free data
Safeguards Privacy, Enhances Accountability remains a challenge… … Additionally, within AI
●Inclusion of Data ethics: establish clear standards, literature there is arange of different definitions of
mandatory training, regular audits, ethics fairness and methods of representing the concept
committees, public accountability mathematically…”
●Contractualism: To act morally is to abide by ●Facts: not decided by how many people believe
principles that no one could reasonably reject them
●Machine bias article: Systematic Unfairness?
Regulatory and Policy Frameworks Feedback Loops? Design Choices? Exacerbate
●Rule of law: nobody is above the law, which treats Disparities?
everybody equally and impartially ● NYC AI action plan: With the release of our AI
●World bank ID4D (Identification for Development): Action Plan, the first-of-its-kind for a major U.S.
support people who lack access (immigrant), city, we are cementing our commitment to this
enabling them to use public services. Everyone emerging technology’s responsible use, and
should have an ID ensuring we are deploying the right tools in the
●Politico Article (Alfred Ng): How or who would be right way
responsible if a data breach happened? The ●Types of justice → ●Distributive justice: also
government point judges as part of addressing known as economic justice, is about fairness in
privacy violations; establish court what people receive, from goods to attention. ●
●What are the methods used to educate and Procedural justice: The principle of fairness is also
engage “ The Public”? Give ID, law enforcement found in the idea of fair play (as opposed to the fair
(regulations) share of distributive justice). ● Restorative justice:
●FOIL/FOIA (Freedom of Information) → there’s a aka corrective justice, The first thing that the
legal/ethical consideration betrayed person may seek from the betrayer is
●Institutional Review Boards (IRB): committees some form of restitution, putting things back as
specific to each institution, but all serve under the they should be. ● Retributive justice: works on the
same ethical principles and legal mandates of principle of punishment, although what constitutes
reviewing research methods involving human fair and proportional punishment is widely
subjects debated. While the intent may be to dissuade the
●Why? There a lot of unethical researchs. Principle: perpetrator or others from future wrongdoing, the
respect for person, beneficence, justice, informed re-offending rate of many criminals indicates the
consent limited success of this approach.
●Level of policies and law
○Int: General Data Protection Regulation ●Montjoye et.al (Unique in the shopping mall: on the
(GDPR) reidentifiability of credit card meta data): Knowing
○Reg: Asia-Pacific Economic Cooperation the price of a transaction increases the risk of
(APEC) Privacy Framework reidentification by 22%; even anonymous, women
○National: India’s Digital Personal Data are more identifiable than men in cc metadata.
Protection Act (DPDP) ●Why matter: equity in services, policy
○Sate/local: California Consumer Privacy Act effectiveness
(CCPA) ●Suggestion: Diverse and representative datasets,
○Sector: HIPAA (Health Insurance Portability Bias detection and auditing, Fairness
and Accountability Act) constraints,Explainability tools, Human oversight
