Academic Research Methodology and AI Community Policy Workshop

AI as a Research and Aggregation Tool

• Efficiency in Search:

  • Traditional research ("in-house") can be difficult as it often requires navigating thousands of websites.
  • AI tools can function like a "clock" or bot that scans thousands of websites per second to find specific items for a project.
  • AI is particularly useful for recommending specific articles or sources for reading.

• Search Engine vs. Information Source:

  • Using AI for research should be viewed as using a high-powered search engine rather than a shortcut to facts.
  • It is used to aggregate data for human review, rather than having the AI "confess" information that the researcher then takes as fact without verification.

The Pedagogy of Research and Potential Learning Loss

• The Concept of Mundane Tasks:

  • There is a risk of losing specific learning experiences when delegating "mundane tasks," such as resource organization, to AI.
  • Learning how one prefers to personally organize resources is a developmental step in the academic process.

• The Cost-Benefit Analysis of AI Use:

  • Students must ask: "What learning am I losing because I used AI and how much of it am I losing?"
  • Delegating tasks like writing or organizing represents a personal sacrifice of learning how to perform those functions.

• The Importance of Slow Research:

  • In high-pressure high school environments, students often seek the fastest possible completion methods to move on to the next task.
  • The classroom goal is to produce knowledge together and emphasize doing things slowly.
  • The process of research, writing, and dissecting articles builds critical thinking and ethical navigation skills essential for life, even for those not becoming historians.

Defining "Mundane Tasks" vs. Unique Intellectual Work

• Unique Human Contribution:

  • True research involves personal, unique knowledge and opinions that lead to a specific intellectual output.
  • Relying too heavily on AI often results in "sloppier" output.

• Appropriate Tasks for AI:

  • AI excels at quantitative, data-heavy tasks such as:
    • Entering data into spreadsheets.
    • Calculating averages.
    • Managing large datasets (e.g., tracking the status of $70,000$ artifacts out of $80,000$ in a museum).

The Economics and Rhetoric of AI

• Market Pressure:

  • AI is currently being aggressively pushed through advertising and billboards as "the future."
  • This push is largely driven by financial interests and people who have money invested in the technology.

• The Myth of Redundancy:

  • The idea that skills like essay writing will become redundant in the future is a narrative meant to sell AI products; it is not a fixed reality.

Non-AI Resources and Traditional Academic Tools

• Search Engine Alternatives:

  • Google remains a viable alternative to AI for many search tasks.
  • SearchWorks (Stanford University Library catalog) is a highly valuable resource that provides different results than Claude or ChatGPT.
  • Developing the skill to search for keywords in SearchWorks and critically select articles is essential.

• Traditional Academic Methodology:

  • Faculty members completed their entire education and academic training, including comprehensive exams, without the use of AI.
  • Performing all research and writing tasks independently is not only possible but the traditional standard of academic excellence.

• Physical Library Benefits:

  • Using resources like the Cecil Green Library allows for physical discovery.
  • Searching the "stacks" or looking at the books physically arranged around a target source on SearchWorks can lead to unexpected and valuable discoveries through indexes and physical browsing.

Environmental and Ethical Implications

• Environmental Impact:

  • Large Language Models (LLMs) and data centers consume enormous quantities of water for cooling.
  • The resources required for AI infrastructure take significant tolls on the land.

• Social Justice:

  • The environmental impacts of AI technology disproportionately affect marginalized people and communities.

Grammar and Accessibility Tools

• Grammarly and Language Support:
  • Grammarly is considered an acceptable AI tool for students.
  • It is particularly helpful for proofreading one's own work, though it must be used in conjunction with a thorough personal reading of the material.
  • For students where English is not their first language, tools for grammar correction are discussed on a case-by-case basis as a cost-benefit decision for clarity versus academic honesty detection risks.

Developing the Community Agreement and AI Policy

• Core Principles of the Policy:

  • Commitment to reflecting on personal learning goals and desired skills.
  • Avoidance of AI to produce work that is turned in for assignments (no AI-generated writing).
  • Commitment to exhausting non-AI resources (Google, SearchWorks, JSTOR, Scholar Labs) before turning to AI.

• Pedagogical Ethic:

  • The classroom is treated as a community of equals where knowledge is co-constructed rather than dictated by the professor.
  • The policy is founded on mutual trust rather than technological policing.

• Trust and Accountability:

  • The instructors choose to trust students rather than requiring surveillance tools like "Track Changes" in Google Docs, though such tools were considered.
  • Breaking the AI policy is viewed as breaking the trust between the student and the learning community.

Technical and Formatting Requirements for Final Projects

• Project Length and Scope:

  • The final project is a $10$-page paper.
  • Footnotes and quotes are included in the page count.
  • The bibliography and title page are NOT included in the $10$-page count.

• Formatting Standards:

  • Font size: $12$-point.
  • Spacing: Double-spaced.
  • Approved Fonts: Standard, easy-to-read fonts such as Times New Roman, Arial, Helvetica, or Georgia.

• Citations and Bibliography:

  • Style: Chicago Style (standard for History).
  • Footnotes are strongly preferred over endnotes.
  • A full bibliography must be included at the end.
  • Every source—whether from a book or the internet—must be cited properly.

• Visual Materials:

  • If using one or two images, they can be included in the text.
  • If using multiple images (e.g., $5$, $6$, or $7$), they should be placed in an appendix to avoid taking up writing space.
  • The core requirement is $10$ pages of substantive writing.

Questions & Discussion

• Question (Audience): Can we use AI to help find specifically focused things, like a specific item for a project?

  • Response (Instructor): Yes, AI can be helpful as a recommendation source or to scan data, but not for synthesizing that data into an argument or taking its facts verbatim.

• Question (Audience): What about citation generators?

  • Response (Instructor): Using citation generators or tools like Zotero is acceptable. Zotero is particularly good and includes features like a Google Docs plugin.

• Question (Audience): Is Grammarly okay?

  • Response (Instructor): Yes, Grammarly is acceptable for proofreading, as long as the student is still thoroughly reading their own work.

• Question (Audience): Are different types of AI better than others?

  • Response (Audience Member): There are many types beyond ChatGPT, such as Claude, Gemini, and Claude Opus. Older generative models (pre-$2022$) mostly predicted the next word and were prone to hallucinations; newer models may have different merits but still should not generate original ideas for the student.

• Question (Audience): Will we be required to use Google Docs with "Track Changes" on?

  • Response (Instructor): While some schools use this to police AI use, the instructors here prefer to rely on verbal agreement and trust, though they recognize it is a standard practice in many high schools.

• Question (Audience): What if I have to write a paper every month? Is $10$ pages a lot?

  • Response (Instructor): While it may feel daunting depending on high school background, a $10$-page paper using $12$-point, double-spaced formatting is a standard academic requirement in college (the "standard across the board").