Organizational AI Policy Proposal Assignment

Jessica McCaughey
George Washington University

This assignment allows professional writing students the chance to consider and make recommendations on the use of Large Language Models (LLMs) in specific workplace contexts. In small groups, students select and research a particular organization and consider the aims, values, internal and external stakeholders, and constraints that could impact the use of LLMs. They research both the larger standards and uses of AI within their chosen industry and the larger business and scholarly conversations taking place as these technologies rapidly evolve. They then develop a policy recommendations proposal for the organization.

Learning Goals

  • Identify and contextualize industry and organization-specific needs and constraints (both rhetorical and practical) as they relate to Large Language Models 
  • Analyze and practice composing in the genre of an internal proposal
  • Collaborate on a large-scale professional research project, including, providing feedback to other team members, and negotiating roles and revisions
  • Critically consider large-scale impacts of Large Language Model use
  • Generate comprehensive, persuasive, and specific recommendations
  •  Anticipate (and ground in research) how the organizations’ needs and actions will evolve along with the technology at hand

Original Assignment Context: Professional Writing course in an Organizational Sciences/Communication Department

Materials Needed: Library resources

Time Frame: Approximately three weeks

Overview:  This assignment, developed for a professional writing class housed in a Communication/Organizational Sciences department, allows students the chance to consider and make recommendations on the use of Large Language Models (LLMs) in specific workplace contexts. In small groups, students select and research a particular organization, and consider the particular aims, values, internal and external stakeholders, and constraints that could impact the use of LLM, and they research both the larger industry standards and uses of AI within it and the larger business and scholarly conversations taking place as these technologies rapidly evolve. They then develop a policy recommendations proposal for the organization.

I’ve taught this assignment only once, and I was very pleased with the outcomes. In their reflection writing, students almost unanimously stated that the assignment had deepened and complicated their understanding of Large Language Models, and they felt prepared to assess, in a more complex manner, the appropriateness and risks of employing the technology in the workplace.

Assignment

In your future careers, you will find that, in addition to evolving *constantly*, artificial intelligence (AI) and the issues related to it will be incredibly industry and organizationally specific. For this assignment, you will work in small groups to develop a generative AI (LLM-specific) policy proposal for an organization of your choice.

Although you should organize your proposal based on your group’s priorities (and the assumed concerns of your chosen organization), your proposal must, at a minimum, address the following concerns:

Rhetorical Concerns

  • Why is an AI policy necessary for this organization?
  • What are the biggest considerations for the organization when it comes to generative artificial intelligence?
  • What are the biggest benefits to generative AI for your particular organization? 
  • What are the biggest risks?
  • What are the values of this organization, and how might the proposals attempt to align these values with engagement of Large Language Models?

Technical and Logistical Concerns

  • Which generative AI platforms, if any, are allowed? Which version, if applicable?
  • If AI use is allowed in some way, should employees cite it? If so, how?
  • Is there a difference for AI use in internal vs. external communications? Or other units or divisions that AI use is acceptable or unacceptable (say, social media)? You might think of categories as required, encouraged, allowed, or forbidden.

Privacy, Property, and Regulations

  • What privacy, data, copyright, intellectual property, or security issues should the organization consider when it comes to AI? How does this proposed policy address these issues?
  • Are there legal standards for AI use in your organization’s particular industry, space, or clientele?

Ethics

  • What are the ethical considerations that the organization should be concerned with when it comes to AI use? These might be broad (all organizations must consider this!) or specific (for our company, X is really crucial for us to grapple with). The policy proposal should grapple with these ethical concerns (using current sources) and make an argument about where the organization is willing to make trade-offs, etc

Training and Evolution

  • How will the policy assure that employees understand the AI they are using and use it responsibly and critically?
  • How do you anticipate that this policy will evolve? Is there a mechanism you can build into the organization’s processes that allows for evolution (of both the technology and the policy)?

Implications

  •  What are the biggest implications of this proposed policy that the organization needs to understand? What are the risks? What are the benefits? What will this mean for the employees? Is this going to eliminate jobs? Create jobs?

Other concerns or sections will be determined by the nature of your particular organization. More than anything else, your proposed policy must be rhetorically sound. By that I mean that it should be very tailored for your particular organization and audience. What would they care about when it comes to this AI discussion? (Efficiency? Being tech-forward? Organizational mission alignment? Other issues?)

Audience

This is a proposal that will need to be approved by members of an internal AI task force. You should assume that on this task force you have high-level colleagues with wildly varying and strong views on large language models. This means that your policy proposal needs to be clear, nuanced, detailed, well-researched, and extremely specific to your organization. You should do everything in your power to anticipate and address counter arguments.

Further, while you will become an expert on these technologies through the course of this project, your imagined audience will have varying degrees of familiarity, so you’re writing must be extremely clear and plain and direct, despite what will be, in many places, it’s very technical nature. 

Breakdown of Major Tasks

  • Create and sign a group contract (in class)
  • Learn about the genre of an organizational policy proposal (in class, we will read, analyze, and evaluate AI policies from a variety of types of organizations)
  • Choose an organization to write your policy proposal for. You may write your policy proposal to any* organization that does not already have a publicly available AI policy. (*The only exceptions include large tech companies.) Some ideas:

o   An advocacy group of some kind

o   A Congressperson or Senator’s office

o   A federal government agency/unit

o   A local government unit

o   A school system

o   A non-profit organization

o   A large or small corporation

o   A local small business

o   An organization within an industry that ties to your major(s) or the interests of the group more broadly

  • Learn as much as you can about your chosen organization
  • Research and read about generative AI broadly, then start researching LLM use in the industry you’ve chosen
  • Decide on a general format for the proposal
  • Draft the proposal, including relevant research
  • Peer review your draft in class: Each student will perform an individual peer review and bring that feedback back to the group; then, you’ll need to assess, negotiate, prioritize, and act on that feedback as you revise
  • Revise
  • Meet with Prof. M. for feedback
  • Revise again (and again, and again!)

Specifics

  • The length of these proposals will vary, but generally we want to aim for 3 – 5 single spaced pages, with bolded headings (and, possibly, subheadings)
  • I believe it would be difficult to successfully support your position with either entirely scholarly resources or entirely popular sources, so you should assume you will use a mix of both; be sure to ensure both credibility and a relevant (recent!) publication date. 
  • You should link to your sources to words or phrases in the document (rather than using in-text citations or footnotes), but you should also have a references section at the end, with APA citations

Presentation

You will pitch your policy proposal to the class as though we are the AI Task Force at your chosen organization.  

Post-Project Reflection

After submitting the group project, please answer the following questions in your individual reflection:

  • How much did you know about the subject of Large Language Models before we started? What was the most interesting/compelling/exciting/disturbing thing you learned about them in the research process?
  • What problems did you encounter while you were working on this proposal? How did you solve them?
  • What kind of feedback did you get from the other groups? In what ways was it useful (or not) to your group?
  • How do you feel about this piece of work? What parts of it do you particularly like? Dislike? Why? What did you enjoy about working on this project? What did you find frustrating about it? 
  • What did you learn about yourself as you worked on this piece?
  • Have any of your beliefs or thoughts about AI changed based on your work here?
  • If you were the teacher, what comments would you make about this piece? What about if you were the task force at the organization? 

Acknowledgements

This assignment was sparked by and developed during a year-long faculty seminar titled “Critical AI & Teaching Writing in the Disciplines” and run by my colleagues Dr. Phyllis Ryder and Dr. Carol Hayes at George Washington University.