4 Chapter 4 Concepts for Ethical Decision Making

Core ethical concepts are central to moral philosophy and ethical decision-making. They each play a distinct but interconnected role in how we evaluate actions, decisions, character, and moral duties. Let’s explore how they relate to ethical decision-making.

1. Good: The concept of good generally refers to what is morally valuable, desirable, or worthy of pursuit. It’s often used to describe outcomes, goals, character traits, or states of affairs.

Key Points:

• It can refer to intrinsic goods (valuable in themselves, like happiness or knowledge) or instrumental goods (valuable because they lead to other goods, like money or education).
• In ethical theories, such as utilitarianism, “the good” is often defined in terms of maximizing happiness or well-being.

Examples:

• Kindness is a good trait.
• Promoting public health is a good goal.
• Pleasure or flourishing is a good outcome.

2. Right: The concept of the right is about moral correctness — what one ought to do. It’s generally used to evaluate actions in terms of their alignment with moral rules, principles, or duties.

Key Points:

• “Right” actions are those that are morally permissible or obligatory.
• Some actions can be right even if they don’t lead to the best possible outcome (depending on the ethical theory).
• Deontological ethics (like Kantianism) emphasize the importance of doing what is right based on duty, not consequences.

Examples:

• Telling the truth is the right thing to do, even if it causes discomfort.
• Returning a lost wallet is a right action based on honesty and respect for others’ property.

3. Wrong: The opposite of right, wrong refers to actions or behaviors that are morally impermissible or reprehensible.

Key Points:

• Wrong actions violate moral duties or principles.
• Something being “wrong” typically carries a moral judgment of blame or disapproval.
• What is considered wrong can vary across cultures or ethical frameworks, but most theories have clear boundaries (e.g., harm to others, dishonesty).

Examples:

• Stealing or lying is usually considered wrong.
• Causing unnecessary harm is morally wrong in almost all ethical systems.

4. Ought: The term ought expresses moral obligation or duty. It tells us what we should do, often regardless of personal desires or consequences.

Key Points:

• “Ought” statements are normative, they prescribe behavior, rather than just describing it.
• Often central to deontological ethics, where certain actions must be done simply because they are morally required.
• It introduces moral imperatives into decision-making (e.g., “You ought to help others”).

Examples:

• We ought to treat others with respect.
• You ought not to cheat on a test, even if you won’t get caught.

How They Relate to Ethical Decision-Making

When faced with a moral choice or dilemma, we engage these concepts to figure out the best course of action. Here’s how they interact in the decision-making process:

1. Evaluating Outcomes – “Good”

• What consequences will result from this action?
• Will it produce more good than harm?
• Helps us evaluate goals, ends, and what we should aim for

Example: A doctor deciding on a treatment that brings about the best patient outcome.

2. Judging Actions – “Right” vs. “Wrong”

• Is this action morally permissible or impermissible?
• Does it violate any moral duties, rules, or rights?
• Assesses if something is morally permissible or obligatory (right)
• Highlights what is impermissible or blameworthy (wrong)

Example: Is it right to lie to protect someone’s feelings?

3. Moral Obligation – “Ought”

• What am I morally required to do here?
• Even if it’s hard, what ought I to choose based on duty or principle?
• Guides what we are morally required to do
• Example: You might not want to apologize, but you ought to if you’ve wronged someone.

AI support in ethical decision-making

• Providing data-driven insights: Helps people weigh outcomes, risks, and benefits more accurately.
• Highlighting overlooked factors: AI can process vast amounts of data and point out implications humans might miss.
• Modeling ethical frameworks: Some systems are designed to mimic ethical theories (e.g., utilitarianism or deontology) to assist in value-based decision-making.

Example in healthcare: AI might help clinicians choose between treatments by predicting long-term outcomes, improving ethical choices around patient welfare.

As a Hidden Influencer (Ethical Nudging or Bias)

AI can unintentionally (or intentionally) nudge decisions in certain directions:

• Recommender systems may shape what options people see first, influencing their perceived “best” choices.
• Algorithmic bias can skew decisions based on flawed training data, reinforcing social inequalities.
• Design decisions (by developers) can bake in values, for example, prioritizing efficiency over empathy.

Ethical concern: If you’re not aware of the AI’s influence, you might make a decision that feels objective but isn’t.

As a Decision-Maker or Recommender

In some systems, AI is not just advising, it’s deciding.

• Triage systems, loan approvals, or parole recommendations may rely on AI to make or support final calls.
• These systems often lack transparency, and their ethical reasoning (if any) isn’t always explainable.

Ethical question: Should AI ever make decisions with moral weight? Who gets to design its ethical compass?

Choosing a Dataset for Research

Define the Objective

• What do you want to achieve?
• You need to be very clear about your research question or project goal.

Example: “Predict the risk of diabetes based on lifestyle factors.”

Determine the Data Requirements

• What kind of data do you need to answer your question?
• Think about:
  • What variables you need (age, blood pressure, genes, etc.)
  • The format (structured tables, images, genomic sequences)
  • Quantity (sample size) and quality (accuracy, completeness)

Explore Data Sources

• Where can you find the data?
• Two big categories:
  • Open data repositories (like Kaggle, UCI Machine Learning Repository)
  • Specialized sources for health and genomics (like NIH databases, TCGA, or MIMIC-III for ICU data)

Evaluate Dataset Suitability

• Does the dataset meet the requirements for your purpose or what you seek to discover?
• Check:
  • Relevance (Does it fit your project?)
  • Size (Enough data?)
  • Quality (Reliable, accurate, and recent?)
  • Metadata (Does it include descriptions of variables?)

Ethical Considerations

• Is it ethical to use this data?
• Especially critical in health/medical fields:
  • Patient privacy (Is the data de-identified?)
  • Proper use according to the data license
  • Consent issues (Was consent obtained for research use?)

Examples of Reputable Datasets for Genomics, Health and Medical, Image, and Physiological Purposes

Genomic Datasets:

• dbGaP (Database of Genotypes and Phenotypes): This database, managed by the National Institutes of Health (NIH), provides data from studies investigating the interaction of genotype and phenotype in humans.
• gnomAD (Genome Aggregation Database): gnomAD offers extensive genomic data, including exome and whole-genome sequences from a diverse range of individuals, enabling researchers to study genetic variation and its impact on health and disease.
• GenBank: This NIH database contains an annotated collection of all publicly available DNA sequences.
• dbVar: The NCBI database of genomic structural variation, including insertions, deletions, and other rearrangements.
• UK Biobank: This large-scale medical research resource provides extensive phenotypical, genetic, and health-record data on over 500,000 participants.

Health and Medical Datasets:

• HealthData.gov: This portal provides access to a wide range of government-produced health datasets.
• CORD-19 (COVID-19 Open Research Dataset): A comprehensive dataset of scholarly articles related to COVID-19, accelerating research efforts during the pandemic.
• National Health Interview Survey (NHIS): This annual, cross-sectional survey provides nationally representative data on health status and utilization in the US.
• World Health Organization (WHO): The WHO’s Global Health Observatory offers various health statistics and data.
• Human Mortality Database (HMD): Provides mortality, population, and other health/demographic data across numerous countries.
• National Center for Health Statistics (NCHS): Offers a wealth of data and tools for health research.
• IQVIA: Provides real-world evidence and health data insights, including electronic health records (EHRs).
• Global Burden of Disease (GBD): A research program that provides estimates of the global health impact of various diseases.
• Institute for Health Metrics and Evaluation (IHME): A global health research center at the University of Washington.

Image Datasets:

• UMIE Datasets: A collection of over a million annotated radiological images, including CT, MRI, and X-ray images.
• OASIS (Open Access Series of Imaging Studies): Neuroimaging data sets of the brain.
• CheXpert: A dataset of chest X-ray images with associated labels.
• Shaip X-Ray Datasets: High-quality X-ray image datasets for research and diagnosis.

Physiological Datasets:

• PhysioNet: This database offers a variety of physiological datasets, including ECG, EEG, and other biomedical signals. It’s a valuable resource for researchers in areas like biometrics, critical care, and more.
• eICU Collaborative Research Database: This dataset focuses on clinical data from intensive care units, providing insights into critical care patients and their management.
• Human Mortality Database: This database is a leading resource for mortality research, offering comprehensive data on mortality rates and trends.
• Other Reputable Datasets: Additionally, the National Center for Health Statistics (NCHS), the Healthcare Cost and Utilization Project (HCUP), and the National Institutes of Health (NIH) databases provide a wealth of health-related data for research.

Final Steps

• Apply the process:
  • You might be tasked with finding a dataset, evaluating it, and justifying why it’s suitable.
  • Sometimes the process will ask you to point out ethical concerns or propose a mini-research project based on the data.
• Finalize and document your choice:
  • Clearly state why you picked a certain dataset.
  • Describe any limitations you foresee.
  • Mention any preparation needed (e.g. cleaning, missing data) before analysis.