Library and ITS Guidance on Artificial Intelligence for Fall 2024

Capabilities and Concerns

Critical evaluation of any AI platform you consider integrating into your teaching or scholarship is key to the ethical utilization of AI. Current generative AI tools can create remarkable content, but have limitations and will often "hallucinate" and create convincing-looking but fake citations or untrue statements. There are serious concerns about bias in the training sets the AI models use, unfair labor practices of the companies behind AI platforms, and the environmental impact of these tools. You should also consider issues of copyright and fair use when designing an assignment that integrates generative AI.

Academic Technology and Swarthmore Libraries are constantly vetting generative AI tools (as well as platforms that integrate AI) and are testing the limits, capabilities, and accessibility of the tools we support. Swarthmore has options for tools that we license which can provide more control. We can help you and your students make an informed decision about when, how, and which tools to experiment with for academic purposes.  

Syllabus statements

We recommend providing clear guidance on acceptable use of AI in your class. This might include a general statement on your syllabus and/or course site, as well as specific policies for individual assignments. Here is a short list of examples that may be helpful as you craft your policies:

  • Three policy examples from Harvard that include a maximally restrictive policy (no AI use), a fully-encouraging policy, and a mixed use policy (see "Policies for the use of AI in courses").
  • Sample statements from across the disciplines curated from the Center for Teaching, Learning, and Innovation at the University of Pennsylvania.
  • Guidance on how to write effective statements from Carleton College.
  • A wealth of resources on writing AI policies and developing sample assignments from the University of Maine.

We also recommend providing guidance on the transparent use of AI tools, such as writing a “statement of use” as part of an assignment or citing the integration of AI generated materials as part of a bibliography. Most disciplinary professional organizations now provide clear citation guidelines and templates, such as the MLA, APA, and Chicago style guides.

Assignments

In some cases you may want to thoughtfully experiment with the use of AI tools in your assignments, but there may be other assignments that you prefer students complete without the aid of these platforms. Some types of assignments are particularly susceptible to the unethical (unintentionally or otherwise) use of generative AI.  ITS and the Libraries can assist you in the creation of dynamic new assignments or the redesign of existing activities to include these considerations. Here are some general recommendations:

Ideas for ensuring students are creating their own work:

  • Requiring a personal reflection or connection to current events
  • Including an oral component or presentation
  • Documenting the process with rough drafts and revisions to track the progress of a student's work
  • Have students use generative AI and critique the output

Alternative assignment ideas:

  • Instead of summarizing a reading, try using social annotation tools such as Perusall or hypothes.is (we can provide you and your students with access and support in using these easy-to-learn tools)
  • Instead of writing a long-format analysis, try a multimodal project (including the creation of videos, podcasts, interactive storytelling, games, etc)
  • Peruse this incredible evolving collection of carefully curated assignments available for remixing and reuse

And remember that you can schedule a conversation with an Academic Technologist and/or a librarian to help you design, implement, and assess these assignments.

Privacy

Most free AI tools ingest user data for training purposes. Submitting student work or requiring students to submit their work to a generative AI tool has privacy risks. Likewise, uploading copyrighted or licensed content may violate legal agreements. We do not recommend entering or uploading any intellectual property, including your own scholarship, research data, student work, and published content, into free AI tools. Again, Swarthmore has licensed platforms through which we can mitigate risks and adjust settings.

If you are planning an AI-based assignment, discuss your ideas with a Librarian or Academic Technologist to evaluate potential use cases to ensure safe use of any AI platform.

Detection (or lack thereof)

We don't believe that there is a tool available that can definitively detect AI use in student work.  We discourage efforts to submit all student work to a system to attempt to determine if AI was used.  

Additional Resources

Contact information

Swarthmore Libraries

Amanda Licastro (Digital Scholarship Librarian)

You can reach out to digitalscholarship@swarthmore.edu or find a area specialist on our website

Academic Technology

  • Ashley Turner (Humanities)
  • Doug Willen (Social Sciences)
  • Andrew Ruether (Natural Sciences and Engineering)

Teaching & Learning Commons

If you’re interested in connecting with other faculty and staff interested in discussion of the impact of generative AI on teaching and learning, the TLC sponsors regular discussions on the subject as part of their Inquiry Group series. Invitations to join upcoming inquiry groups are sent out in the TLC newsletter, but feel free to contact us any time.