Ethel is an experimental tool designed to support students in understanding the course material.
It operates using retrieval-augmented generation (RAG) and large language models (LLMs) to generate responses.
The information provided by the Ethel is for general informational purposes only and should not be considered authoritative, accurate, or complete.
By using Ethel, you acknowledge and agree to the following disclaimers and limitations.
Disclaimers
- No Guarantee of Accuracy. While Ethel aims to provide relevant and helpful information, it is subject to several limitations, including: potential inaccuracies, outdated information, or hallucinations, that is, fabricated details that are not based on factual data; misinterpretation of complex or ambiguous queries; lack of full comprehension of nuanced topics; failure to properly attribute sources or reference materials.
We strongly advise you to verify its responses through official course materials, textbooks, and the course assistants. ETH and the course staff do not guarantee the correctness, reliability, or completeness of the Ethel’s responses.
-
No Legal, Academic, or Professional Liability or Warranty.
ETH, the course staff, and the developers of Ethel are not responsible for any misunderstandings, errors, or decisions made based on Ethel’s responses.
Ethel does not provide legally binding advice, guidance, or official rulings.
Ethel’s responses should not be used as a sole source for assignments.
Any reliance on the responses of Ethel is at the user’s own risk.
Finally, Ethel is provided “as is” without any warranty, express or implied. ETH and its affiliates disclaim all liability for damages, losses, or consequences arising from the use of this assistant. By using this tool, you acknowledge and accept these risks.
-
User Responsibility.
Users of Ethel are responsible for cross-referencing responses with the course material, consulting instructors or subject-matter experts for clarification, and avoiding reliance on Ethel for graded assignments or assessments.
-
Updates and Modifications.
ETH, the course staff, and the developers of Ethel reserve the right to modify or discontinue Ethel at any time without notice. These disclaimers and limitations may be updated periodically to reflect new findings or system improvements.
Known Failure Cases
-
Ethel is conditioned on course material from last year’s course iteration, not this year.
Hence, its references and information may be incorrect, and it may not be able to answer questions correctly about, for example, project deadlines, grading information, or the timeline and dates.
-
Ethel may generate incorrect, misleading, or entirely false information.
-
Ethel is actively conditioned on the course material, and thus it may give unexpected answers to off-topic questions by trying to contextualize the query in the context of the course content.
The bot may misinterpret questions, especially when queries are vague or highly specific.
-
As an AI model trained on diverse datasets, Ethel may reflect biases inherent in its data sources.
-
Even if course materials are explicitly included in the retrieval augmentation system, Ethel may still generate responses based on general knowledge rather than course-specific details, which can diverge from the way the material is taught in the class.
