Google Gemini and NotebookLM

  1. What is Gemini?
  2. Who can use Gemini?
  3. How do I use Gemini?
    1. Prompts
    2. Good prompts and bad prompts: an illustration using Gemini
  4. What is NotebookLM?
  5. Who can use NotebookLM?
  6. Data protection and considerations for Gemini and NotebookLM at UH
  7. Limitations and concerns
  8. Where can I learn more about Gemini, NotebookLM, and AI at UH?
    1. Gemini
    2. NotebookLM
    3. AI at UH
  9. Get help for Gemini and NotebookLM

What is Gemini?

Gemini is Google’s advanced generative artificial intelligence (AI) system that can understand and generate human-like text, translate languages, write different kinds of creative content, and answer your questions in an informative way. It can help you with tasks like writing emails, summarizing articles, reviewing code, and creating stories or poems. Gemini can even generate images based on your input. Essentially, it's like having a super-intelligent assistant that can help you with a wide range of tasks.

Gemini leverages the power of large language models (LLMs), a type of machine learning (ML) focused on language processing. LLMs are trained to predict and match patterns in human language using massive amounts of text, code, and other written data. Generative AIs like Gemini use that training to generate content for you by predicting the desired output to your prompts.

Who can use Gemini?

Faculty and Staff: All faculty and staff at UH can use Gemini as a Core Google Workspace App.

Students: Students at UH who have enrolled in Google@UH Consumer Apps will be able to use Gemini.

How do I use Gemini?

Prompts

Gemini responds to queries that you construct and provide called prompts. You can also upload images or record audio as part of your prompt for Gemini to evaluate and respond to. The results generated by Gemini are called responses. When submitting a prompt, consider that Gemini cannot read your mind and absent specifics, constraints, or examples, Gemini may interpret its response to your prompt in a variety of ways. To be a more effective prompt writer and in turn receive better and more accurate responses from Gemini, consider the following tips:

  • Be Specific and Clear:
    • Avoid ambiguity: Instead of "Write a story," try "Write a short science fiction story about a robot who falls in love with a human."
    • Provide context: If your prompt relates to a specific topic, provide background information or relevant details.
  • Use Keywords Effectively:
    • Include relevant keywords: This helps Gemini understand the desired output. For example, "Write a persuasive essay on the importance of renewable energy."
    • Use precise language: Instead of "good," use "concise," "engaging," or "informative."
  • Give Instructions:
    • Include relevant keywords: This helps Gemini understand the desired output. For example, "Write a persuasive essay on the importance of renewable energy."
    • Use precise language: Instead of "good," use "concise," "engaging," or "informative."
  • Provide Examples:
    • Show, don't just tell: If you want Gemini to write in a particular style, provide a short example.
    • Use few-shot learning: Give Gemini a few examples of the type of output you're looking for.
  • Iterate and Refine:
    • Experiment with different prompts: Try rephrasing your prompt to see how it affects the output.
    • Refine based on results: Adjust your prompt based on Gemini's initial response.

Good prompts and bad prompts: an illustration using Gemini

Consider this prompt and response:

Create an image of a pair of sandals on a tropical beach.

Sure, here is an image of a pair of sandals on a tropical beach.

pair of sandals floating on top of a beach

Gemini did indeed create an image that looks like a pair of sandals on a tropical beach. That’s pretty neat, but something looks a little off. The sandals appear to be floating on top of the beach. Large and central to the framing of the image, they are covering up most other elements in the image’s background. Due to my poorly constructed and unspecific prompt, the image Gemini created wasn’t anything close to what I had imagined, but from Gemini’s perspective, it was exactly what I had asked for. If a picture is worth a thousand words, it would stand to reason that only a handful of words may not make for an interesting picture.

Let’s try again. Consider this new prompt:

Create an image of a sketch of a tropical beach. A pair of sandals lie on the sand as waves lap close by. Palm trees on the beach sway in the breeze. The afternoon sun peeks through clouds that hug mountains in the background.

Sure, here is an image of a sketch of a tropical beach.

pair of sandals on a beach with waves, palm trees, mountains, and clouds in the background

This time, I was more specific about the style of image I was imagining (a sketch). I provided additional details about the various content I’m looking for in the image (sandals, waves, palm trees, mountains, and clouds). I even provided information about time of day and lighting (afternoon sun). The result? An image much closer to what I had imagined.

What is NotebookLM

NotebookLM is an AI-powered research and note-taking tool developed by Google Labs. NotebookLM leverages Gemini to help transform how users interact and engage with information by providing intelligent and personalized synthesis and analysis of that information. Users can upload sources, which are documents such as text documents, PDFs, web URLs, YouTube videos, or audio files for NotebookLM to synthesize and analyze.

NotebookLM is designed to intelligently analyze content you provide to:

  • Generate summaries: Quickly summarize key points from uploaded documents.
  • Provide explanations: Offer concise and informative explanations of complex topics.
  • Answer questions: Respond to user inquiries based on the content of their uploaded files.

NotebookLM can help users save time and improve their workflow and enhance productivity by:

  • Creating documents: Assisting in the creation of various document types, such as timelines, FAQs, and study guides.
  • Refining content: Suggesting improvements to existing documents, such as better phrasing or more concise wording.

NotebookLM aims to transform how users engage with and make sense of their information by providing an intelligent and personalized assistant that helps them analyze, synthesize, and create content.

Who can use NotebookLM?

As a Google@UH Consumer App, NotebookLM is available to all staff, faculty, or students who have enrolled in Google@UH Consumer Apps.

Data protection and considerations for Gemini and NotebookLM at UH

Data protections for Gemini and NotebookLM at UH

Faculty and staff use of Gemini as a Core App is governed by the master agreement between UH and Google. Student use of Gemini and all users’ use of NotebookLM, as Consumer Apps, are subject to Google’s Terms of Service.

Regardless of your affiliation with the University, when interacting with Gemini or NotebookLM with your Google@UH account, data from your Gemini or NotebookLM activity will never be reviewed by humans nor used by Google to train or improve their AI models.

Data considerations, guidance, and best practices when using generative AI at UH

While data from your Gemini activity will never be reviewed by humans nor used by Google to train or improve their AI models (when used with your Google@UH account), you should continue to exercise caution when interacting with and providing prompts or data to any generative AI (genAI) models, including Gemini and you should avoid supplying sensitive information, including personal, confidential, or proprietary information or data. When interacting with any generative AI, ITS recommends that you:

  • do not include any PII (personal identifiable information) in your prompts
  • do not include student work or FERPA protected data in your prompts
  • do not include other sensitive, confidential, or proprietary information in your prompts

The University of Hawaii has initiated a comprehensive, cross-functional effort to equip the UH community with essential tools, guidance, and best practices for effectively integrating AI into academic and professional environments. Additionally, efforts are underway to develop and implement policy changes that address the evolving impact of AI on work and education.

UH community members are encouraged to view the working draft of AI Tools UH Guidance and Best Practices for Faculty and Staff.

  • UH community members may send questions and comments regarding the draft AI Tools UH Guidance and Best Practices for Faculty and Staff to datagov@hawaii.edu.

Stay tuned to ITS’ AI information and resources page for updates!

Limitations and concerns

Accuracy of information and “hallucinations”: Generative AI models can confidently generate factually incorrect or misleading information, often presented in a convincing manner. Always evaluate information generated by AI for credibility. Gemini offers a feature to “double-check” its response.

Data bias: Generative AI models are trained on massive datasets. If these datasets reflect existing societal biases (e.g., gender, race, socioeconomic status), the AI models may perpetuate and even amplify those biases in their responses.

Lack of transparency: Many generative AI models are complex and opaque, making it difficult to understand how they arrive at their outputs. This lack of transparency can raise concerns about accountability and the potential for unintended consequences.

Ethical issues & copyright: Determining ownership of AI-generated content and ensuring fair compensation for creators can be challenging.

Google continually seeks to mitigate these issues in Gemini and their other AI tools by adhering to their AI Principles.

Where can I learn more about Gemini, NotebookLM, and AI at UH?

Gemini

NotebookLM

AI at UH

Get help for Gemini and NotebookLM