Skip to main content

Honors Option

Co-developed by D. Caballero & V. Sawtelle

  • Eligibility: A 3.0 or better in CMSE 101, since the honors option builds on the course learning objectives rather than replacing them.
  • Students: Individual

Overview#

The honors option for CMSE 101 is about understanding the real costs of Artificial Intelligence infrastructure, energy, water, and human labor, set against the hype that surrounds it, and about developing your own critically pragmatic response. It is not a physics or engineering deep-dive into how computers generate heat; the emphasis stays on the same Data-Tools-Practices-Actions lens the rest of the course uses, applied to the resource and labor footprint of AI systems.

Critical pragmatism means holding two things at once: naming what is exploitative, wasteful, or oversold about a system, while also taking seriously that critique alone does not tell anyone what to do next. Your project should not stop at documenting harm or hype. It should end with something usable, a policy recommendation, a design constraint, or practical guidance for people in your own space (your campus, your major, your household), that responds to what you found, and that someone could reasonably argue against.

For this honors option you will research at least one of the topics below, tracking the inputs and outputs, documenting the energy and water requirements of the systems you study, and identifying whose labor makes the system possible and who is exposed to its harms.

Choose a Topic#

  1. AI policy, economics, and power. The political economy of AI, examined through the lens of “who is in control?” Pick one of the four angles below and go deep on it, rather than surveying all four:

    1. Policy. A specific current or proposed regulation of how AI systems are built, trained, or deployed: what it requires, who wrote it, and whose interests it protects or exposes.
    2. Energy policy. A specific government or utility decision responding to AI’s electricity demand (a rate case, a permitting decision, a subsidy): who requested it, who approved it, and who pays either way.
    3. Tokenomics. The business economics of running AI models: what a specific provider charges per token, whether that price reflects the actual cost of running the model, and who profits from the gap.
    4. Education report and policy. The MIT report on AI and education (or a comparable report), and how a specific institution or policy has responded to its findings.

    Ground your “who is in control?” analysis in one of the following (they take different views of the same question, so pick the one that fits your angle best):

    • P. Marx, Tech Won’t Save Us (podcast) — see the episode already listed under “Sources for energy” below.
    • E. Zitron, Better Offline (newsletter/podcast) — critiques of AI-industry hype and power concentration.
    • G. Durán, The Nerd Reich: Silicon Valley Fascism and the War on Democracy, Avid Reader Press (2026).

  2. Cases of research in energy/water use. Pick one documented case (a company, a specific data center, a region) and quantify its energy and/or water footprint using primary reporting or peer-reviewed data, not press releases.

  3. Open and closed systems. Compare one open-source/open-weight AI system to one closed, proprietary system on a specific dimension (energy transparency, who can audit it, who can build on it), and argue which approach serves the public better on that dimension.

  4. Sources for energy. Pick one energy source currently used to power AI infrastructure (e.g., diesel or gas-turbine generators, nuclear, grid electricity) and trace who decided to use it, what environmental or community impact it has had, and why cleaner alternatives weren’t chosen.

  5. Edge computing. Compare running a specific AI task on-device (a phone, an IoT chip) versus in the cloud, and weigh the energy, privacy, and cost tradeoffs of each for that task.

  6. Tokens as a measure of AI energy efficiency. Evaluate whether “tokens per watt-hour” or a similar metric actually measures what it claims to, using at least one benchmark or leaderboard as your evidence.

  7. Labor behind the systems. Pick one stage of the AI supply chain (data annotation/content moderation, mineral extraction, data center construction/maintenance, e-waste recycling), and document who does the work, under what conditions, and what protections (if any) exist for them.

Every topic: two required lenses#

Whichever topic you choose, your write-up must address both of the following, not just the resource numbers.

1. Hype vs. evidence. Find at least one specific industry or marketing claim about your topic (a company promise, a projection, a press release) and hold it up against independent evidence. Where does the claim hold up, and where does it not?

2. Labor and exploitation. Name at least one group of people whose labor makes your topic’s systems possible, and at least one group exposed to their harms. These are often the same people (for example, the workers building or cooling a data center are rarely the ones who benefit most from it). Be specific about who they are, where they work, and what evidence documents their conditions.

Optional research direction: futurist ideologies#

If your topic touches on how AI’s future is talked about (this fits topic 1’s “who is in control?” question especially well), one way to dig into the hype lens is to ask who benefits from imagining today’s costs as temporary. TESCREAL (Transhumanism, Extropianism, Singularitarianism, Cosmism, Rationalism, Effective Altruism, Longtermism) is Timnit Gebru and Émile Torres’s term for a bundle of ideologies that treat problems like energy use, labor conditions, or inequality as acceptable now because a sufficiently advanced future (faster computers, off-world expansion, a frozen-and-revived humanity) is assumed to make them moot.

If you take this direction, don’t just describe what each ideology believes. Pick the one or two that are actually relevant to your topic and make an argued case: who is invoking this kind of thinking about your topic, what are they using it to justify or dismiss, and what do you think of that move? This is a research direction, not a required checklist for every project, and it needs your own commentary to be worth including.

Your proposal#

Documenting the problem is not enough. Close your project with a specific, critically pragmatic response: a policy recommendation, an institutional guideline, a design constraint, or practical guidance for people in your own space (your campus, your major, your household, your own AI use). It should be specific enough that a reader could disagree with it, and it should respond directly to what your hype and labor analysis found. Neither assume the technology will fix itself nor assume nothing can be done.

What do I turn in for the project?#

Project Format#

Your final product will be a write-up in a style that is scholarly research. Your benchmark is a 3000-word essay, a 10-minute video with slides, or a scientific poster. Your research must be well-documented, include at least one table with calculations of energy requirements and/or costs, and must include your proposal (see above).

Supporting Assignments#

  • You will need to submit a detailed plan, including your question or topic selected from the list above. Include your final project submission type, your schedule, and 2-3 sources you have selected to help you complete the work. Once this has been accepted by Danny, you may submit your honors option to the appropriate honors-college paperwork.

  • You will submit a mid-project prototype or outline, due no later than mid-November. Turn in either:

    • A cover letter describing your work and accomplishments so far, including changes to your initial plan, challenges you faced, and how you are feeling about the project, or
    • A detailed outline of your planned project. The detailed outline should organize your research findings and include quotes and citations.

Final Product#

Your final product is due to the dropbox on D2L no later than December 21, 2026, though you may choose an earlier deadline. If you choose the presentation or poster style, you must also submit a video recording of your presentation.

Resources to get you started#

Cases of research in energy/water use

Open and closed systems

Sources for energy

Edge computing

Tokens as a measure of AI energy efficiency

Hype vs. evidence

Labor and exploitation

Data for energy and cost calculations