Generative AI Tools For Education – Gemini AI (1)

The role of AI in education can not be denied. As AI becomes commonplace across industries, educators can leverage its potential to enhance the teaching and learning process. This post series looks into practical Generative AI tools applicable in educational settings. We have curated a continuously updated list of Generative AI tools, while emphasising that AI serves as a valuable supplement but not a replacement of educators. The main function of AI tools is to assist with tasks and streamline process, not replacing competent educators nor eliminating jobs. AI serves as a powerful tool to streamline tasks and personalise learning.

Furthermore, using generative AI tools like Google Gemini AI requires minimal technical expertise. Through prompts, users can guide the AI to generate desired educational content. Refining prompts leads to progressively better results, while advanced techniques like prompt engineering can further optimise outcomes. This post showcase how to leverage Google Gemini AI through sample frameworks for generating educational content.

Gemini AI

1. Gemini: Gemini is the new name for Google Bard. It is available freely for use and subscriptions to gain access to advanced features. It is available in over 40 languages and more than 230 Countries and territories. It is available via the website and currently being rolled out for Mobile on Andriod and Apple iOS. Gemini is a multimodal language model, which means it can understand and process information from different modalities, such as text, code, audio, images, and video. This allows it to perform a wider range of tasks than traditional language models that are only trained on text data. For example, Gemini can:

• Generate different creative text formats, like poems, code, scripts, musical pieces, email, letters, etc. from a single prompt.
• Answer your questions in an informative way, even if they are open ended, challenging, or strange.
• Generate different responses based on the context of the conversation.
• Translate languages.
• Write different kinds of creative content.

The free version of Gemini is Gemini Pro1.0 while the Advanced version is Gemini Ultra 1.0

No special technical skills or prompt engineering expertise are required to use generative AI. These tools use natural language prompting where, you simply ask for what you want done and keep refining until you get your expected output. However, you can start by following these points to guide your prompting:

• Be specific about your learning goal.
• Provide context
• Use clear and concise language.
• Ask open ended questions.
• Leverage different prompts.
• Refine your prompt based on my response.

The points above were from prompting Gemini as follows:

Prompt:

How do I prompt you to get the best output in the context of Education?

Output:

1. Be specific about your learning goal:

• Are you looking for factual information on a topic (e.g., “Explain the life cycle of a butterfly”)?
• Do you need help understanding a concept (e.g., “Compare and contrast photosynthesis and cellular respiration”)?
• Are you practicing a skill (e.g., “Write a persuasive essay arguing for the importance of space exploration”)?

The more specific you are, the better I can tailor my response to your needs.

2. Provide context:

If your question relates to a specific class or assignment, mention the subject and any relevant details. This will help me understand the background knowledge you have and provide a more focused response.

The clearer your prompt, the easier it is for me to understand what you’re looking for. Avoid overly complex sentences or jargon.

4. Ask open ended questions:

Don’t just ask for yes/no answers. Open ended questions encourage deeper exploration and learning. (e.g., Instead of “What is the capital of France?”, ask “What factors influenced the selection of Paris as the capital of France?”)

5. Leverage different prompts:

I can handle various prompts besides questions. Here are some ideas:

• Summarize a complex topic (e.g., “Summarize the key ideas of Darwin’s theory of evolution”).
• Create a study guide for a specific subject (e.g., “Create a study guide for the upcoming history exam on the American Revolution”).
• Brainstorm creative writing prompts related to a particular theme (e.g., “Generate writing prompts about artificial intelligence”).

6. Refine your prompt based on my response:

My first response might not be exactly what you need. Don’t hesitate to rephrase your prompt based on my initial answer to get a more refined output.

Similarly, you can follow a formula given in acronyms (SCRIBE) learnt from the “AI in Education summit ” that took place on Saturday 27th of April 2024.

SCRIBE

S – Specify: State a role you want the generative AI to play e.g (You are an experienced Educator preparing a lesson plan for your learners in their first year at University)

C – Context: Provide relevant information you would like the generative AI put into consideration for the output e-g (Most of these learners are new to the subject area)

R – Responsibility: Outline the tasks you want the generative AI to perform.

I – Instructions: Give detailed step by step for the generative AI to follow.

B – Banter: Ask follow up questions to refine the output.

E – Evaluate: Ask the generative AI to ask you to assess the quality and accuracy of the prompt use.

The Five “S” Framework

Another tool you can use as a guide to prompting for Education purpose is the Five “S” framework developed by AI for Education

S – Set the scene: Provide contexts in terms of roles, expertise or environment to guide the expected output e-g (You are an expert STEM instructional designer and teacher)

S – Specific: Be specific in the instruction, clearly stating the tasks and details of what you need to be included in the output e-g(Use 5E model to create a 60 minutes lesson in a subject area)

S – Simplify: Simplify your language in a conversational mode, avoiding jargons or technical terms e.g (Create an engaging lesson plan that aligns with the national curriculum in year 6 English language).

S – Structure: Structure the output by instructing the generative AI to generate the output in specific format, for a specific audience or environment e.g (Create the output in a tablular format, generate the output in a simplified English for learners of English as a foreign language)

S – Share: Share feedback with the generative AI on the output as you would do in a conversation for instance if you need a different format or a revision of the output to meet your expected output or exploration of different options e.g( I would like you to revise this output in an informal conversational tone, I would like you to create the output in a table format)

Sample Prompting for Gemini AI

Now that we have looked into various frameworks and guidelines for prompting generative AI, we can use sample prompts to test out Gemini AI. Note that all the guidelines and frameworks will work on other generative AI with little or no tweaking as they all work on the same general principle of generative output based on input of text prompts. However the output may be different based on the source of the information being generated, which has more to do with the model used.

For Educational purpose, there are already prompt libraries and resources freely available for use. Check out the reference section for links. You can use these prompt templates as starter based on your need, while you continue to build on it till you get your required output. However, you can always come up with your own prompt based on the framework shared above or just start as you would prefer to, without using framework. The more your practice, the more you learn how to get your required output and you can also come up with your own framework specific to your needs.

Prompt template to Create Course Module outline and outcome

I am teaching a 14-week college-level {CourseName} course. For this course I am using a learning management system and want to have an online module for each week of the course. Suggest a 14-week module outline for me that includes the title of the module and a few of the topics that will be covered and write those in the form of module level learning outcomes. I know you will create a great flow of module topics and respective module outcomes that are specific and measurable. Remember, you are an expert in assessment and know that understand is not a measurable outcome.

The above prompt template is from Alchemy library of prompts for Higher Education, to create a course module outline and outcome. If we analyse the prompt template, we can see that it follows guideline provided by Gemini AI when asked for prompting for the best output and it also aligns to both of the given frameworks. Back to Gemini AI with prompt to analyse the above prompt template we get the following:

Prompt:

{The text of the prompt template}. Analyse this prompt template in terms of the advise you gave on prompting you for the best output?

Output:

Your prompt template analyses well based on the advice I gave for getting the best educational output:

• Specificity: You clearly state the course level (college), format (online modules), and desired outcome (14-week module outline).
• Context: You mention using a learning management system, implying the course has an online component.
  Clarity: The prompt uses concise language and avoids jargon.
• Open Endedness: You ask for suggestions, not a specific answer, allowing for creative response.
• Refinement Potential: You acknowledge the need for refinement based on the suggested outline.

Areas for Improvement:

Missing Information: The most crucial missing information is the course name ({CourseName}). Knowing the subject allows for tailoring the module titles and learning outcomes to the specific field.

Overall, your prompt is excellent!

Here’s how I can further enhance it based on the missing information:

Improved Prompt:

“I am teaching a 14-week college-level {CourseName}course. For this course I am using a learning management system and want to have an online module for each week of the course. Could you suggest a 14-week module outline with titles and a few learning outcomes for each module? The learning outcomes should be specific, measurable, and action-oriented, avoiding terms like ‘understand’.”

By including the course name, I can provide a more relevant and focused module outline with specific and measurable learning outcomes tailored to your subject.

In the generated output, Gemini AI indicated clearly that the prompt template is excellent but also highlight the area of improvement, which is the name of course in question. The place holder ensures that the template can be customised, by simply replacing it with a specific course name. Gemini AI also compare the prompt template with each of its suggested points for prompting for the best output. Additionally Gemini AI gave an improved prompt template, which we will use with a specific course name in this example.

Prompt:

“I am teaching a 14-week college-level Software Engineering with Python course. For this course I am using a learning management system and want to have an online module for each week of the course. Could you suggest a 14-week module outline with titles and a few learning outcomes for each module? The learning outcomes should be specific, measurable, and action-oriented, avoiding terms like ‘understand’.”

Output:

14-Week Software Engineering with Python Course Outline

This outline suggests a structure for your online modules with specific and measurable learning outcomes for each week.

Note: This is a general framework. You can adjust the content and depth based on your specific curriculum and student needs.

Week 1: Introduction to Software Engineering & Python Programming

• Module Title: Welcome to Software Engineering with Python!
• Learning Outcomes:
  • Students will be able to define software engineering principles and their application in problem-solving (written assignment).
  • Students will be able to set up a Python development environment and run basic code examples (in-class activity).

Week 2: Python Fundamentals – Variables, Data Types, & Operators

• Module Title: Building Blocks of Python: Variables, Data Types, & Operators
• Learning Outcomes:
  • Students will be able to create and manipulate variables of different data types (in-class coding exercise).
  • Students will be able to apply various arithmetic, comparison, and logical operators to evaluate expressions (coding quiz).

Week 3: Control Flow & Conditional Statements

• Module Title: Making Decisions: Control Flow with Conditional Statements
• Learning Outcomes:
  • Students will be able to write Python code using `if`, `else`, and `elif` statements to control program flow (coding assignment).
  • Students will be able to use conditional statements to solve simple decision-making problems in Python (coding challenge).

Week 4: Loops & Iteration

• Module Title: Repetition Made Easy: Loops in Python
• Learning Outcomes:
  • Students will be able to implement `for` and `while` loops for iterating through sequences in Python (hands-on coding session).
  • Students will be able to write Python code using loops to automate repetitive tasks (programming project).

Week 5: Functions & Code Reusability

• Module Title: Building Reusable Code with Functions
• Learning Outcomes:
  • Students will be able to define and call functions in Python to modularize code (coding practice).
  • Students will be able to design functions with parameters and return values for efficient code reuse (programming assignment).

Week 6: Introduction to Data Structures: Lists, Tuples, & Dictionaries

• Module Title: Organizing Data Effectively: Lists, Tuples, & Dictionaries
• Learning Outcomes:
  • Students will be able to create and manipulate lists, tuples, and dictionaries in Python for data storage (coding workshop).
  • Students will be able to choose appropriate data structures based on the nature of the data they are working with (written analysis).

Week 7: Data Manipulation with Strings and Files

• Module Title: Working with Text & Files in Python
• Learning Outcomes:
  • Students will be able to perform string manipulation operations in Python using built-in functions and methods (coding exercises).
  • Students will be able to read and write data to/from text files using Python (programming project).

Week 8: Midterm Exam (Review of covered concepts)

Week 9: Introduction to Object-Oriented Programming (OOP) Concepts

• Module Title: Object-Oriented Programming Fundamentals
• Learning Outcomes:
  • Students will be able to define the core concepts of OOP (classes, objects, inheritance, polymorphism) (written explanation).
  • Students will be able to create basic classes and objects in Python (coding activity).

Week 10: Designing Classes and Object Interactions

• Module Title: Building Classes & Object-Oriented Applications
• Learning Outcomes:
  • Students will be able to design and implement classes with attributes and methods in Python (coding project).
  • Students will be able to demonstrate the principles of inheritance and polymorphism using Python code (programming challenge).

Week 11: Exception Handling & Error Management

• Module Title: Handling Errors Gracefully: Exception Handling
• Learning Outcomes:
  • Students will be able to identify and handle common exceptions using `try`, `except`, and `finally` blocks in Python (coding practice).
  • Students will be able to write robust Python code that anticipates and gracefully handles potential errors (programming assignment).

Week 12: Introduction to Testing & Debugging

• Module Title: Ensuring Code Quality: Testing & Debugging
• Learning Outcomes:
  • Students will be able to explain the importance of unit testing in software development (written reflection).
  • Students will be able to use a Python testing framework (like `unittest`) to write unit tests (coding workshop).

Week 13: Building a Software Project

• Module Title: Putting it All Together: Project Development
• Learning Outcomes:
  • Students will be able to apply the learned concepts to design, develop, and document a software project (course project proposal and presentation).
  • Students will be able to demonstrate the principles of software engineering practices (version control, collaboration) using Git and GitHub (coding assignment with Git integration).

Week 14: Course Wrap-up & Assessment

• Module Title: Software Engineering with Python: Looking Ahead
• Learning Outcomes:
  • Students will be able to reflect on their learning experience and identify areas for further development (written self-assessment).
  • Students will be able to complete a comprehensive final exam that assesses their understanding of key software engineering concepts and their application with Python (final exam).

This outline provides a foundation for your 14-week Software Engineering with Python course. Remember to adapt the content based on the specific learning objectives you have for your students. You can also include additional modules to cover more advanced topics like web development, databases, or specific Python libraries depending on your curriculum.

With the revised prompt, Gemini AI generated, a 14-week course outline with students learning outcome indicators for each week. From the outline generated, we can continue to prompt Gemini AI for contents for each topic, practical assignments for students engagement, learning assessments aligned to learning objectives, feedback and much more.

In the second part of this post, we explore co-creating a prompt framework with Gemini AI. We use this framework to generate content for week one of the outlined course above. Similar to earlier interactions, we can directly query Gemini AI for prompting guidance. As an example, we previously asked “How do I prompt you for the best educational output?” This prompted Gemini to provide a helpful guide. We then requested it to analyse the template we used based on its own guidance, resulting in a thoughtful response. This collaborative approach empowers educators to refine prompts for optimal results. Following this approach, in part two of this post, we ask Gemini AI to design a framework for generating educational content.

In line with our workflow process, we have researched, edited and refined some of the contents with a combination of generative AI tools, Gemini AI and copilot.