Applied Artificial Intelligence
Currently catalogued as Adv Artificial Intelligence
Advanced AI is no longer studied chiefly as a set of standalone models; in industry it now ships as autonomous, tool-calling agents that plan, retrieve, and act, so the practitioner's work has shifted toward specification, instrumentation, evaluation, and accountability for behavior no human authored line by line. The course therefore keeps its classical content on probabilistic reasoning, planning, natural language processing, robotic navigation, and deep learning while adding agent architectures, evaluation harnesses, guardrails, and provenance as graded practice.
Current catalog prerequisites — (CSC 415 or 3400).
Current description → proposed description
This course builds upon the foundation developed by CSC 3400 by introducing students to a variety of cutting-edge artificial intelligence applications and systems. Included are such topics as probabilistic reasoning, planning and decision-making, natural language processing, robotic navigation, and deep learning. In addition, the ethical and moral implications of advanced artificial intelligence systems will be examined from both scientific and philosophical perspectives. The limitations of artificial intelligence will be examined via the framework of the Epistemological Assumption and a Christian worldview
Prerequisites: (CSC 415 or 3400).
This course builds upon the foundation developed in CSC 3400 by introducing students to a variety of cutting-edge artificial intelligence applications and systems. Included are such topics as probabilistic reasoning, planning and decision-making, natural language processing, robotic navigation, and deep learning. In addition, the ethical and moral implications of advanced artificial intelligence systems will be examined from both scientific and philosophical perspectives. The limitations of artificial intelligence will be examined via the framework of the Epistemological Assumption and a Christian worldview. Because such systems now reach industry as autonomous, tool-calling agents, students will also build, instrument, and evaluate agent architectures that plan, retrieve, and act under human oversight, taking responsibility for verifying machine-generated work they did not write themselves.
AppliedAIF2F_Proposal.xlsx), where it appears as “Applied Artificial Intelligence”.
Workbook lists this as a Course Change BCI (CSC 4410).
What changes
- Retitled from "Adv" to "Applied" Artificial Intelligence per the workbook BCI
- Adds agent architectures: planning, retrieval-augmented generation, tool-calling under human oversight
- Formal evaluation harnesses, LLM-as-judge rubrics, and failure-mode error analysis replace demo-only assessment
- Deployment guardrails: prompt injection, tool supply-chain risk, human-in-the-loop escalation, cost and latency budgets
- Provenance and verification of machine-generated work become explicit graded deliverables
6 proposed outcomes, mapped to 15 program outcomes
Each outcome below is written to be observable and assessable, and each is mapped to the program learning outcomes for which it produces evidence.
Students will be able to implement an advanced artificial intelligence application that integrates probabilistic reasoning, automated planning, and deep learning with a retrieval-augmented, tool-calling agent architecture to address an authentic stakeholder problem.
PLO 4.2 is supported because building an integrated agent architecture for a complex problem is the proposing half of that outcome's propose-and-evaluate requirement, which CLO 2 completes by evaluating the same system; PLO 3.2 is supported because the CLO requires the system to serve an authentic stakeholder problem rather than a benchmark dataset, which is the human-centered application work that outcome names, and CLO 4 carries the human-oversight and governance half of the same outcome so that it is not assessed on this element alone.
Students will be able to evaluate the accuracy, robustness, and failure modes of an agentic AI system using task-level benchmarks, LLM-as-judge rubrics, and structured error analysis, reporting what the measured outcomes indicate about deployment readiness and about the harm the system's failures would cause to the people it serves.
PLO 4.1 is supported in both of its halves, because benchmark results and structured error analysis are the critical analysis of data-driven outcomes and the required judgment about the harm the system's failures would cause to the people it serves is the ethical and societal implication that outcome also demands; PLO 4.2 is supported because benchmarking, LLM-as-judge review, and failure-mode analysis of the system proposed in CLO 1 is the evaluating half of that outcome's propose-and-evaluate requirement; PLO 6.2 is supported because benchmark scores supply the statistical insight while LLM-as-judge rubrics and error analysis supply the qualitative insight that outcome pairs, both applied to reach a conclusion about deployment readiness.
Students will be able to critique, through the framework of the Epistemological Assumption and a Lutheran Christian worldview, both the epistemic and moral limits of advanced artificial intelligence, including deep-learning opacity, hallucination, encoded cultural bias, and delegated autonomous decision-making, and their own design decisions and problem-solving approach, stating what they would do differently in light of that critique.
PLO 6.3 is supported because critiquing encoded bias and the moral limits of delegated autonomous decision-making is a direct assessment of AI developments for fairness and moral impact; PLO 4.3 is supported because the CLO turns that critique on the students' own design decisions and problem-solving approach and requires them to state what they would do differently, carried out within the Lutheran Christian framework the CLO names, which is precisely the faith-informed self-reflection that outcome defines; PLO 2.2 is supported specifically by the encoded cultural bias element, which requires students to recognize that a system can fail across the cultures and perspectives it was not built to respect.
Students will be able to design the guardrails and stewardship controls for a deployed autonomous agent, including prompt-injection and tool supply-chain defenses, human-in-the-loop escalation thresholds, and compute, cost, and latency budgets, for a system serving a community or nonprofit partner.
PLO 2.3 is supported by the explicit compute, cost, and latency budgets together with the safety controls, which are the ethical resource management and responsible-usage practices that outcome names; PLO 2.1 is supported because the CLO situates the deployment in service of a community or nonprofit partner rather than an abstract client; PLO 3.2 is supported because setting human-in-the-loop escalation thresholds is human-centered application design by definition, and choosing the stewardship controls and budgets under which a partner's live system may act is a governance and human-factors judgment applied alongside the engineering rather than a purely technical one.
Students will be able to construct, within an interdisciplinary team, an applied artificial intelligence solution that reconciles competing technical, legal, ethical, and theological constraints on a real stakeholder problem.
PLO 3.1 is supported because the CLO requires students to reconcile technical (data science), ethical, and theological constraints on a single problem, which is the ethics, data science, and theology combination that outcome names, and the workbook marks this PLO as assessed at exit in this course; PLO 3.3 is supported by the requirement that the work be produced within an interdisciplinary team balancing the technical and ethical dimensions of the solution; PLO 5.2 is supported because reconciling competing constraints requires students to navigate diverse professional and theological views inside that team.
Students will be able to articulate the design, the evaluation evidence, and the provenance of an artificial intelligence system, including which artifacts were machine-generated and how each was verified, in a written technical report and in a presentation for a non-specialist audience.
PLO 5.1 is supported because the CLO requires both written and spoken formats and requires disclosure of what was machine-generated and how it was verified, which is the transparency and integrity that outcome names; PLO 5.3 is supported by the presentation for a non-specialist audience, which is the translation of technical AI knowledge for lay listeners; PLO 6.1 is supported because the CLO requires students to explain what the system's evaluation evidence means to an audience outside the field, which is the interpretation and explanation of AI data outcomes that fosters understanding of AI's impact across fields.
Program outcomes this course reaches
Filled cells are program learning outcomes with at least one supporting course learning outcome in this course. Sparse coverage is expected — no single course carries all eighteen.
The workbook maps this course too
The proposal workbook's assessment map already assigns program learning outcomes to this course. The outcomes above were written to cover it.
| Program outcome | Workbook level | In this draft |
|---|---|---|
| PLO 2.1 Service through Technology | D | covered |
| PLO 2.2 Engagement with Diversity | D | covered |
| PLO 2.3 Stewardship of AI Resources | D | covered |
| PLO 3.1 Interdisciplinary Problem-Solving | AE | covered |
| PLO 3.2 Real-World AI Applications | I,D,AE | covered |
| PLO 3.3 Collaboration in AI Projects | I,D,AE | covered |
| PLO 4.1 Data and Ethical Analysis | I,D,AE | covered |
| PLO 4.2 Innovative AI Solutions | I,D,AE | covered |
| PLO 4.3 Reflective Decision-Making | I,D,AE | covered |
| PLO 5.1 Clear and Ethical Communication | I,D | covered |
| PLO 5.2 Interpersonal Skills in AI Collaboration | I,D | covered |
| PLO 5.3 Engagement with Broader Audiences | I | covered |
| PLO 6.1 Data Interpretation in AI | I,D,AE | covered |
| PLO 6.2 Quantitative and Qualitative Analysis | I,D,AE | covered |
| PLO 6.3 Critical Ethical Analysis | I,D | covered |
I = Introduced · D = Developed · AE = Assessed at Exit