Proceed with caution when using generative AI tools (including within platforms such as Slack, Tableau, Google, and Office365 that are rapidly adopting them).
While recognizing the potential for innovation and creativity that generative AI tools offer, a subject that will be explored in a variety of ways on campus, the guidance that we offer below focuses largely on mitigating risks, some of which carry serious consequences, that we wish to bring to your attention.
Many don’t offer robust privacy controls. Unless the College’s licensing agreement ensures content isn't used externally or to train algorithms without anonymization, please follow the guidelines below:
- In interactions with these tools, do not use information that you; the College; our faculty, staff, or students would not want shared publicly. Examples include financial statements, payroll or salary details, disciplinary records, College processes, and system architecture and/or vulnerabilities.
- Do not use personally identifiable information (PII) or data files that are not anonymized, such as donor history, student transcripts, tenure review details, resumes or application letters, payment history, etc.
- If you are considering licensing a new tool or utilizing a new AI feature in a tool you already use, please work with AskIT to ensure that the tools and services you procure or are using on behalf of the College have appropriate privacy and security protections and are assessed for risk prior to use.
- If you have already begun the use of new tools or features, configure the platform to not use or retain chat/prompt history, if possible. Once data are placed into these platforms, there are no guarantees that they can later be removed from them.
- Artificial intelligence tools are advancing rapidly, but they can still be inaccurate and occasionally produce hallucinations. Moreover, they may exhibit inherent biases. Always verify the suggestions or information they provide, particularly in technical disciplines and matters related to student work and the honor code.
If you have questions that focus on the technical aspects of AI, please reach out to David Hamilton, chief information officer. For all other questions, please contact the co-chairs of the task force, Chris Kingston and Jaya Kannan.
We thank you in advance for your attention to these matters and look forward to navigating this new frontier with the Amherst community.