Syllabus

  1. Syllabus
    1. Important details
    2. Course description
    3. Learning outcomes
    4. Prerequisites
    5. Class format
    6. Time expectations
    7. Communication
    8. Assessment
      1. Attendance
      2. In-class discussion participation
      3. Reading quizzes
      4. Paper presentation and discussion leading
      5. Hands-on activities
    9. Devices in lecture
    10. Course Schedule
    11. Academic integrity
    12. Generative AI policy
    13. Audio-video recording policy
    14. Accommodations
    15. University policies, academic support, student support, and campus emergency information
    16. A note on self care

Important details

Course description

Generative Artificial Intelligence (GenAI) tools are rapidly changing how software is built, maintained, and reviewed. This seminar explores emerging topics in large language model (LLM)-based agentic software engineering, including: foundations of agentic development tools, agentic applications across the software development lifecycle, and methods for evaluating their effectiveness and impact on sociotechnical software systems in practice. The course combines paper reading, presentation, and discussion. This seminar is designed for students interested in exploring agentic software engineering; no prior experience with LLM-based tools is required.

Learning outcomes

After taking this course, among others, students should be able to:

  • Compose a synthesized position on research about GenAI development tools by appraising the design, evidence, and conclusions of that research, and communicate it to others
  • Evaluate new claims about GenAI development tools as they emerge, distinguishing demonstrated capability from marketing
  • Explain what GenAI development tools are, how they are being used in the software industry, and where their capabilities and limitations lie in practice
  • Create an assessment approach for determining whether a given GenAI development tool improves a specific, measurable outcome for a particular use case, selecting appropriate methods and justifying the choice
  • Apply knowledge of the GenAI development tools landscape to describe the range of applications of GenAI across software development and maintenance, including code generation, code review, program repair, testing, and security
  • Analyze the design of an agentic development tool in terms of its core components and the function each serves

Prerequisites

This course does not have formal prerequisites, but we describe background knowledge that will help you be successful in the course. In a nutshell, we expect basic exposure to GenAI tools and basic programming skills, but do not require software engineering experience.

Class format

Classes will be in person, every Monday 3:30 PM to 6:00 PM in ROME 204.

Readings. There will be reading materials for each lecture, which students are required to read through before the class. Short quizzes on the assigned readings will be given in class without advance notice.

Interactive activities. There will be interactive activities interspersed through the lectures. These will include active learning activities on course content and short quizzes.

Paper presentations and discussions. Each meeting centers on discussion of the assigned readings. One student leads the presentation and discussion of each paper, and everyone is expected to arrive ready to contribute to said discussion.

Hands-on activities. Hands-on activities are completed during class over the course of the semester. These involve working directly with GenAI development tools and with the kinds of evidence used to evaluate them.

Lectures. A few meetings open with a short lecture where background is needed before discussion, including the first class, the meetings on evaluation, and the introduction to agents.

Time expectations

  • 2.5 hours of direct instruction (i.e., class time) per week
  • 5 hours of independent learning (i.e., out of class time) per week
  • 112.5 hours total per semester

Communication

The course uses Blackboard for grading, announcements, and supplementary documents. Discussion and questions will be managed on Slack. Slides, assignments and the schedule will be posted on this website.

Please use Slack for discussion and questions, including clarifying assignments. We prefer you write us on Slack instead of email. The instructors hold weekly office hours. If you cannot make it to office hours, contact us via Slack and we will find an alternative time to meet.

Assessment

Component Weight
Attendance 10%
In-class discussion participation 20%
Reading quizzes 30%
Paper presentation and discussion leading 20%
Hands-on activities 20%

Attendance

We consider in-class participation an integral part of the learning experience. A sign-in sheet will be at the front of the room each week. The sign-in sheet is how attendance is recorded, and you are responsible for signing in each class. Two absences are waived, no questions asked. Everyone is also counted present for the first meeting regardless of whether they attended, to accommodate add/drop.

Your attendance grade is calculated as the number of meetings you attended plus two, divided by the total number of meetings, capped at 100%. For example, with 14 meetings, attending 12 earns 100%. Attending 10 earns 86%.

If you cannot attend class due to a medical issue, family emergency, interview, or other unforeseeable reason, please contact us about possible accommodations. We try to be as flexible as we can, but will handle these cases individually.

In-class discussion participation

We strongly believe in in-class discussions and in-class exercises and want all students to participate, e.g., answering or asking questions in class, sharing own experiences, presenting results, or participating in in-class votes and surveys. We will give many opportunities for participation in every lecture. Since this is a relatively small class, the quality of the course will hinge on the quality of student engagement and participation.

Engagement includes preparation, openness to learning, active listening, note taking, and other forms of reflection on course material and discussion. Participation involves contributing to class discussion in ways that advance the learning goals of the course. We note student participation with in-class activities to include as a component in grading. We will provide feedback at mid-semester so that you can check in on how you’re doing. Again, please talk to us if you need accommodations.

We assign participation grades as follows:

Grade Criterion
100% Participates actively at least once in most lectures (4 lectures waived, no questions asked)
90% Participates actively at least once in two thirds of the lectures
75% Participates actively at least once in over half of the lectures
50% Participates actively at least once in one quarter of the lectures
20% Participates actively at least once in at least 3 lectures
0% Participation in less than 3 lectures

Reading quizzes

Reading quizzes are short, closed-note, and administered in class without advance notice. Each quiz consists of two or three questions drawn directly from the reading assigned for that meeting and takes about five minutes. Quizzes are designed to be straightforward for a student who has read the assigned material and difficult for a student who has not. They are not designed to test recall of fine detail.

There will be at least eight quizzes over the semester, and your two lowest quiz scores will be dropped no questions asked. This covers an off day, a difficult reading, or a class you missed. Because quizzes are unannounced, they cannot be made up; the two dropped scores exist to absorb this.

Paper presentation and discussion leading

Each student will be assigned one paper that they are responsible for leading the presentation and discussion of. Each paper session should last roughly 50 minutes, and follow the structure outlined in the Presentation Guidance and Structure document. You are responsible for sending us your slides via email by noon on the Thursday before your session. Five points will be deducted from your presentation grade for each day the slides are late, up to a maximum of 20 points. The rubric is also available.

When you are not presenting, you are still expected to have read the paper and to come with something to say. A discussion leader can only work with what the room brings.

Hands-on activities

Hands-on activities will be completed and submitted during class over the course of the semester, there will be at least four. These involve working directly with GenAI development tools and with the kinds of evidence used to evaluate them. Activity dates are announced in advance so you know when to bring a laptop. Because these are done in class, they cannot be made up, but your lowest score is dropped.

Devices in lecture

Research shows that using devices on non-class related activities harms both the device user’s learning, and other students’ learning as well. Therefore, in general, we do not allow the use of devices during class. If you genuinely use your laptop for class-related activities (note-taking, etc), tell us, and we will make an exception. However, we ask that if you do so, you are careful to keep your devices in note-taking mode (and don’t stray to email, homework, etc). In addition, you will be required to sit in the back row of the lecture to minimize the impact your screen has on others. Hands-on activity days are an exception. Those four meetings require a laptop, and the dates will be announced in advance so you know when to bring one.

Because this class is a paper-reading seminar, you will need the readings in front of you on paper. We will hand out printed copies of the following week’s papers at the end of each class, so you should never need to print anything yourself. Annotate them however you like; they are yours to keep. PDFs are also posted on Blackboard if you prefer to read digitally before class or want a searchable copy.

Course Schedule

Printed copies of each paper are handed out at the end of the preceding class, and PDFs are linked below and can be found in the Box paper repository. This page is the authoritative version of the course schedule and is updated as the semester goes.

Date Topic Readings Hands-On Activity
Mon Aug 24 Course Structure + What is a GenAI Development Tool? None (welcome to class)  
Mon Aug 31 GenAI Development Tool Use in the Software Industry Butler et al., Dear Diary: A Randomized Controlled Trial of Generative AI Coding Tools in the Workplace, 2025.

Dell’Acqua et al., Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality, 2023.
Activity 1
Mon Sep 14 Unanticipated Risks, Downstream Consequences, and Ethical Implications of Adoption Choudhuri et al., Why Johnny Can’t Think: GenAI’s Impacts on Cognitive Engagement, 2026.

Hasan and Biswas, What Breaks When LLMs Code? Characterizing Operational Safety Failures of Agentic Code Assistants, 2026.
 
Mon Sep 21 Anatomy of Agents Pt 1: Context, MCP, Skills Yang et al., SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering, 2024. (pages 1-10 only)

Rombaut, Inside the Scaffold: A Source-Code Taxonomy of Coding Agent Architectures, 2026. (pages 1-35 only)
Activity 2
Mon Sep 28 Anatomy of Agents Pt 2: Harness and Loop Engineering Yang et al., Better Harnesses, Smaller Models: Building 90% Cheaper Agents via Automated Harness Adaptation, 2026.

Madaan et al., Self-Refine: Iterative Refinement with Self-Feedback, 2023.
 
Mon Oct 05 Multi-Agent Systems and Autonomous Agentic Workflows Cemri et al., Why Do Multi-Agent LLM Systems Fail?, 2026.

Qian et al., ChatDev: Communicative Agents for Software Development, 2024.
Activity 3
Mon Oct 19 Assessing Tooling and Judging Performance Claims Pt 1: Human-Centric Evaluations Xia and Miller, Do These Violent Delights Have Violent Ends? Measuring the Post-Merge Fate of Agentic Code, 2026.

Becker et al., Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity, 2025. (pages 1-12 only)

Becker et al., We Are Changing Our Developer Productivity Experiment Design, 2026.
 
Mon Oct 26 Assessing Tooling and Judging Performance Claims Pt 2: Tool-Centric Evaluations Position: Coding Benchmarks Are Misaligned with Agentic Software Engineering, 2026.

Evtikhiev et al., Out of the BLEU: How Should We Assess Quality of the Code Generation Models?, 2023.
Activity 4
Mon Nov 02 Code Generation and Implementation He et al., Speed at the Cost of Quality: How Cursor AI Increases Short-Term Velocity and Long-Term Complexity in Open-Source Projects, 2026.

Liu et al., Debt Behind the AI Boom: A Large-Scale Empirical Study of AI-Generated Code in the Wild, 2026.
 
Mon Nov 09 Code Review and Quality Assurance Adams et al., Automating Low-Risk Code Review at Meta: RADAR, Risk Calibration, and Review Efficiency, 2026.

Alami and Ernst, Human and Machine: How Software Engineers Perceive and Engage with AI-Assisted Code Reviews Compared to Their Peers, 2025.
 
Mon Nov 16 Testing Jain and Le Goues, TestForge: Feedback-Driven, Agentic Test Suite Generation, 2025.

Alshahwan et al., Automated Unit Test Improvement Using Large Language Models at Meta, 2024.
 
Mon Nov 30 Program Repair and Refactoring Xia and Zhang, Automated Program Repair via Conversation: Fixing 162 out of 337 Bugs for $0.42 Each Using ChatGPT, 2024.

Yang et al., Revisiting Unnaturalness for Automated Program Repair in the Era of Large Language Models, 2025.
 
Mon Dec 07 Software Security and Software Supply Chains Spracklen et al., We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs, 2025.

Liu et al., Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale, 2026.

Ghanem, Builder, Defender, Breaker: The Case Against Removing the Human from the AI-Driven Security Lifecycle, 2026.
 
Wed Dec 09 (Designated Monday) Unanticipated Risks, Downstream Consequences, and Ethical Implications of Adoption Pt 2 + Parting Message Afroz et al., “AI Slop is DDoSing Open Source”: Understanding the Impact of AI-Generated Contributions on Open Source Sustainability, 2026.

Motwani et al., Secret Collusion Among AI Agents: Multi-Agent Deception via Steganography, 2024.
 

Note: the final meeting is Wednesday December 9, a designated Monday.

Hands-on activity dates are announced in advance. Bring a laptop on those days.

Academic integrity

The University’s Code of Academic Integrity applies in full to this course. The work you present in this course must be your own. If your presentation includes text, figures, data, or ideas that came from somewhere other than the assigned paper, cite the source. This includes other papers, blog posts, talks, and any figure or illustration you did not create yourself. If you are paraphrasing an argument you encountered elsewhere, say where it came from. Using existing material without proper citation is plagiarism, a form of cheating.

Discussing the readings with your classmates is encouraged and is a normal part of how research communities work. Preparing a presentation for a paper you were not assigned to lead, or having someone else (or something else) prepare yours, is not. If you are unsure whether something is permitted, ask in advance. The minimum penalty for a violation is a zero on the presentation component, and violations will be reported through University channels, which may carry additional consequences under the University’s Policy on Academic Integrity.

Generative AI policy

Let me be straightforward about why this policy exists. A presentation you generated rather than made teaches you nothing and wastes the class’s time. The point of the exercise of creating the presentation is the thinking, and the thinking is the part a model cannot do for you.

Please do not submit AI slop, meaning raw or low-quality output from GenAI tools. You are responsible for all content you submit, and submissions will be evaluated based on the substantive content. You are free to discuss your work with other people and with AI systems, but the work that you submit must reflect your own thinking and intellectual labor.

Using GenAI to draft a presentation or reading quiz response would violate this policy. Using GenAI to transform bullet points into coherent prose would also violate this policy. If you create the content, in the format of the assignment, and then use GenAI to improve the grammar of prose you have already written, that is permitted, so long as you document how you used the system. You may use GenAI to create illustrations for slides but the source of the illustration must be properly documented in the presentation.

Audio-video recording policy

Recording class sessions in any audio or video format is expressly prohibited without the written consent of the instructor. When recordings are permitted, they must be used for individual, educational use only and are covered under the Family Educational Rights and Privacy Act (FERPA) and must not be shared with anyone outside your course-section.

Accommodations

If you wish to request an accommodation due to a documented disability, please contact Disability Support Services at dss@gwu.edu. If you have an accommodations letter from the Disability Support Services office, we encourage you to discuss your accommodations and needs with us as early in the semester as possible. We will work with you to ensure that accommodations are provided as appropriate.

University policies, academic support, student support, and campus emergency information

Information on University policies, academic support, support for students in and outside the classroom, and GW campus emergency information can be found at bulletin.gwu.edu/university-syllabus-policies/.

A note on self care

Please take care of yourself. Do your best to maintain a healthy lifestyle this semester by eating well, exercising, avoiding drugs and alcohol, getting enough sleep and taking some time to relax. This will help you achieve your goals and cope with stress. All of us benefit from support during times of struggle. You are not alone. There are many helpful resources available on campus and an important part of the college experience is learning how to ask for help. Asking for support sooner rather than later is often helpful.

If you or anyone you know experiences any academic stress, difficult life events, or feelings like anxiety or depression, we strongly encourage you to seek support. Counseling and Psychological Services (CAPS) is here to help: call 202-994-5300 (24/7) and visit their website. Consider reaching out to a friend, faculty or family member you trust for help getting connected to the support that can help.


Courtney Miller, George Washington University.

This site uses Just the Docs, a documentation theme for Jekyll.