Williams

AI Ethics — AI and Information Literacy

• 发布单位Publishing unit Williams College Libraries
• 官方来源Official source https://libguides.williams.edu/c.php?g=1470913&p=11044275
• 页面日期Page date 2026-07-26

AI and Privacy

Artificial intelligence is built on data - vast amounts of data. Some of that data is yours. Ideally, each individual determines for themself how, when, and by whom their data is used. In reality, much of that data is scraped from the web without notice or consent. Or collected in security breaches and sold. Furthermore, user data is often collected and used in the training and refining of generative AI models. Different tools have different privacy policies, which often vary based on the level of access a user has purchased. We encourage you to find and read the privacy policies of any tools you plan to use and to be careful about the information that you share with them.^{数据与隐私Data & privacy} Avoid sharing someone else's intellectual property or copyrighted material with these tools.^{知识产权IP & copyright} If you aren't sure, you can always ask a librarian

Local AI: a private option

Small, locally run LLMs offer us the chance to experiment and explore with different models while maintaining a safe and secure environment. Software like LM Studio allows you to download and run LLMs on your device, without your data ever leaving your computer. Here at Williams, Ephbot, a local LLM running on a Raspberry Pi is available for checkout from the Libraries' Equipment Loan Center.^{机构工具Institutional tools} See below for more information

AI and Intellectual Property

Generative AI engages with questions of intellectual property and copyright both at the point of input and of output. Our current large language models have been trained on a corpus of work scraped from the internet, without the authors' permission. Visual tools such as Stable Diffusion and Midjourney are trained on the works of countless artists, also without consent or compensation. Some have argued that the companies developing these AI tools have infringed on their copyright protections and have sued on those grounds.

On the output side, questions of authority, authorship, and citation of genAI content continue to evolve.

AI and Academic Honesty

There are many AI tools that you may be considering adding into your academic workflows. Before doing so, double check your proposed use against your professor's AI policies, the assignment requirements, and the Honor Code^{课程规则优先Instructor override}. What may be acceptable use in one class may be in direct violation of another professor's policies.^{摘录Excerpt}

Thoughtful and intentional engagement with these tools includes working to understand their inherent limitations and strengths, considering how their output is informed by their training and which points of view they may fail to consider, weighing their wider societal and environmental impacts, taking appropriate measures to protect your privacy and comply with intellectual property laws, and always verifying their output.^{核实责任Verification duty}

OIT's current guidance on generative AI detectors.

Citing AI

The conversation around the citation of AI tools is still developing, though a few scholarly style manuals have released guidance. Click around below for that guidance as well as some arguments for and against citing AI tools in scholarly research.

Documenting your use of AI

Because each new interaction with an LLM (or a research assistant tool built on an LLM) produces a novel result, replicating your work is often difficult. Below are some tools for documenting your work with AI.^{过程留证Process evidence}

A browser extension that allows you to easily save and share your generative AI interactions.

Using the "Save page now" feature, you can archive your chatbot discussions using the URL.