MediaArtTutorials

CMST 2ZP3

Glitching tools - AI data poisoning

---

W8: Conduct Individual Research

Research-Creation Project · Glitching tools - AI data poisoning

Groups of 3–4 students

Objective

Students will begin individual research on AI data poisoning and its artistic application in glitch art. This research establishes the critical foundation for your group’s glitch-based creative project. Each student will explore one of four key themes by identifying a specific angle of interest, selecting and annotating relevant sources, and drafting guiding research questions.

AI data poisoning is a technique in which manipulated or malicious data is deliberately introduced into a training dataset to corrupt an AI model’s behavior. This tactic can be used to subvert surveillance systems, critique algorithmic bias, or explore creative disruption.

Each student will explore one of four key themes:

• Glitch Aesthetics and Machine Error as Art
• Algorithmic Bias and the Ethics of Data Corruption
• Poetics of Prompting: Language, Contradiction, and AI Interpretation
• Speculative Media and Machine Perception

They will define a specific research angle, select and annotate relevant sources, and draft guiding research questions.

---

Materials Required

• Digital or physical notebook for organizing citations and ideas
• Laptop or device for research

---

Activities

Complete the following in order. Ask your professor or TA for help as needed.

---

[5 minutes] Topic Selection

With your group, divide the four themes so that each student focuses on one:

• Glitch Aesthetics and Machine Error as Art

  Explore how visual glitches, distortions, and digital errors can be embraced as creative tools. Consider how artists use these breakdowns to critique perfection, destabilize meaning, or represent rupture.

• Algorithmic Bias and the Ethics of Data Corruption

  Investigate how data poisoning reveals or disrupts biases within AI systems. Reflect on the ethical implications of using corruption intentionally—whether as resistance, critique, or subversion.

• Poetics of Prompting: Language, Contradiction, and AI Interpretation

  Examine how the phrasing of prompts shapes image generation. Focus on how contradiction, ambiguity, or poetic inputs can mislead or confuse the AI, turning language into a glitch-producing tool.

• Speculative Media and Machine Perception

  Study how machines “see” and generate images through training data. Consider how artists can manipulate or disrupt these processes to explore speculative futures, alternative visions, or synthetic realities.

Once themes are chosen, students work independently for the remainder of this activity.

---

[30 minutes] Research Angle + Source Planning

Open a document (Word or Google Docs) and Identify and Skim Potential Sources:

Academic Sources (3–5 options):

• Use McMaster Library databases or Google Scholar
• Search for scholarly articles, chapters, or papers
• Suggested search terms: "AI data poisoning" AND ethics, "machine learning" AND "artistic manipulation", "AI bias" AND aesthetics OR art

Public Sources (5–8 options):
Look for reliable public-facing media such as:

• News articles (e.g., CBC, The Guardian, Wired, The New Yorker)
• Podcasts (e.g., Code Switch, IRL Podcast by Mozilla)
• TED Talks (search on ted.com)
• Critical essays or blog posts by artists, activists, or cultural critics

Skim headlines, intros, and section titles — look for relevance, not depth (yet)

---

[30 minutes] Select Final Sources + Annotated Bibliography

Choose:

• 1 academic source
• 2 public sources

Create an annotated bibliography:

• Cite each source in APA format
  • APA Citation Guide
• Write a 3–5 sentence summary of each source’s key ideas
• Explain how each connects to your topic or storytelling concept

---

[25 minutes] Draft 2–3 Research Questions

Based on your research, write 2–3 open-ended questions that will guide your thinking and creative work.

These should help you:

• Analyze your topic from cultural, political, or aesthetic perspectives
• Reflect on your own relationship to the theme (identity, body, memory, tech, etc.)

Sample Questions:

• How can corrupted visual data function as a critique of surveillance capitalism?
• In what ways do glitch aesthetics reflect digital decay, failure, or resistance?
• What are the ethical stakes of using adversarial inputs in art and design?

Strong research questions should not only engage with the technical and ethical dimensions of AI, but also open up space for artistic intervention and experimentation.

---

Submit Partial (Individual) Research Report

You submission document must have:

• Full Name + Student Number
• Theme Title (e.g., Algorithmic Bias and the Ethics of Data Corruption)

Research Notes:

• Brief description of your chosen research angle
• 2–3 guiding research questions

Annotated Bibliography:

• 1 academic source + 2 public sources
• APA citations
• 3–5 sentence annotations per source

➡️ Export your file as a PDF
📄 Filename: PartialResearchReport-Lastname.pdf

---

Before the Next Class

• Read your three sources thoroughly
• Highlight key arguments, case studies, or examples
• Gather quotes, notes, or visuals that inspire you

---

📤 Submission

Type	File Name	Who Submits
Individual Research Report	PartialResearchReport-Lastname.pdf	Each student

⚠️ Follow the submission protocols carefully. Incorrect submissions may result in lost points.

---