Public Syllabus Summary

Machine Learning: Ethics and Society (Fall 2026)

This dated summary follows the course syllabus revised October 9, 2026. That document is labeled STOR 323; this website retains its existing STOR 323I course page. Use the course website for current information and Canvas for enrolled-student announcements, submissions, and grades. Announced adjustments take precedence over this summary.

Course Description

The course studies ML and AI as socio-technical systems: data, models, objectives, interfaces, human judgment, institutional incentives, and public consequences. Its central question is when a system creates benefit or harm, and where responsibility enters its design and use.

Four connected modules integrate technical analysis with ethical reasoning:

  1. Foundations and ethical issues: supervised learning, representation and generation, moral philosophy, bias, fairness, and privacy.
  2. Learning while acting: online learning, regret, A/B tests, bandits, exploration costs, and long-term welfare.
  3. People, platforms, and rules: strategic responses, games and equilibria, information design, mechanism design, and content ecosystems.
  4. Generative AI and society: sustainability, alignment, hallucination, reasoning and agents, safety, creativity, and learning.

Students should have basic probability, statistics, calculus, linear algebra, and programming experience in Python or R. Prior ML coursework is not required.

Learning Goals

  • Explain how core ML systems work and identify the assumptions connecting learned models to deployed decisions.
  • Apply ethical frameworks to concrete questions of fairness, representation, privacy, autonomy, welfare, and responsibility.
  • Analyze dynamic and strategic systems using online learning, game theory, welfare, and mechanism design.
  • Evaluate the mechanics and consequences of language models, reasoning models, and agents.
  • Develop evidence-based technical and ethical arguments, address objections, and propose feasible interventions.

The technical core includes reading formulas, calculating small examples, tracing algorithms, and explaining the consequences of assumptions. Deeper work may take the form of a derivation, a coding experiment, or an empirical/system analysis. Advanced real analysis and training large neural networks from scratch are not required.

Materials and Purchases

No required purchases, paid textbook, API subscription, or commercial software license. Required readings, notes, cases, and code are supplied through course materials or free public sources. Students need Python/Jupyter/Colab or an equivalent R environment, numerical and plotting libraries, and a spreadsheet tool for small audits.

Assessment and Participation

ComponentWeight
Five unannounced pop quizzes15%, 3% each
Five homework assignments20%, 4% each
In-class exams40%
Final group project25%: presentation 10%, report 15%

Attendance and speaking in every discussion are not directly graded. Substantive discussion contributions may earn up to five additional percentage points, at the instructor’s discretion.

For mock-exam score M and course-exam score E, the exam component is max(0.40M + 0.60E, E). The mock exam cannot lower the component below the course-exam score.

The final project is completed in groups of two or three and includes both conceptual analysis and a substantive technical component. Each group gives a 15-minute presentation and submits a report of no more than eight pages.

GradePercentage
A / A-93-100 / 90 to below 93
B+ / B / B-87 to below 90 / 83 to below 87 / 80 to below 83
C+ / C / C-77 to below 80 / 73 to below 77 / 70 to below 73
D+ / D / F67 to below 70 / 60 to below 67 / below 60

Course Policies

Late work. Students may use two 48-hour extensions on eligible individual assignments without giving a reason. These do not apply to in-class activities, the in-class course exam, group presentations, or the final examination period. Contact the instructor early if more time is needed; announced due-date adjustments are authoritative in Canvas.

Collaboration. Discussion, study groups, and peer feedback are encouraged. Submitted work must make each student’s own reasoning visible. Unless an assignment says otherwise, discussing concepts and debugging is allowed, but submitting another person’s prose, code, calculations, or argument as one’s own is not. Group projects include a contribution statement.

Generative AI. Tools may support brainstorming, coding, source discovery, and revision unless an assignment states otherwise. Students may not outsource the central reasoning being assessed. Disclose the tool, task, retained/rejected output, and how claims, citations, reasoning, and code were independently checked. Students remain responsible for submitted work.

Diverse scholarly perspectives. This course engages diverse scholarly perspectives to develop critical thinking, analysis, and debate. Inclusion of a reading, case, argument, or other course material does not imply endorsement by the instructor or the University.

Dated Calendar Anchors

The October 9 syllabus schedules the mock exam for October 28 and the in-class course exam for November 18. Final presentations are planned for November 20, 23, 30, and December 2; the report is due December 2 at 11:59 p.m. These are dated syllabus plans, not a check of subsequent announcements.

The syllabus is a course guide, not an express or implied contract. Changes are announced through Canvas. For academic integrity, accommodations, and student support, consult the course syllabus and the relevant UNC resources.

docs