Getting Started with Copilot – Learning Collaborative (2024)

AI Literacy Day – April 19, 2024
Jerry Slezak, Director, Digital Learning Support

http://learn.umw.edu/copilot

Artificial Intelligence (AI): the ability to make decisions by operating on explicit instructions and predefined rules. Traditional AI is not designed to create original content from existing data like generative AI can.

Generative AI: a type of AI that is trained on data and can generate novel content, such as text, images, music, and code. It continuously learns and refines its outputs based on user interactions.

Chatbot: a software application or web interface that is designed to mimic human conversation through text or voice interactions.

Large Language Model (LLM): a specialized type of artificial intelligence that has been trained on vast amounts of text to understand existing content and generate original content.

LLM Training: the process of teaching LLMs to understand and generate human language. This is achieved through feeding the model massive amounts of text, image, or audio data, and then using algorithms to learn patterns and predict what comes next. LLMsgenerate responses based on patterns they’ve learned from the data, not from any form of understanding or intent.

Prompts: a command, question, request, or statement that is used to direct an AI toolto generate the response, text, or image you want.

Prompt Engineering: crafting questions or instructions (prompts) to guide AI toolsin creating more accurate and desirable responses.

There are different types of AI GPTs that can be used for different purposes. This includes not just text generation, but also creation of things like images, music, video, audio, and computer code.

1. Microsoft Copilot: Known for its sourcing abilities, free internet access, and advanced LLM model.It uses OpenAI’s most advanced LLM.

2. ChatGPT: The original AI chatbot that gained worldwide recognition.It’s known for its ability to perform a wide range of tasks.

3. Anthrophic’s Claude: This AI chatbot excels at summarizing documents, making it a great tool for digesting large amounts of information quickly.

4. Perplexity.ai: This chatbot is best for prompt ideation, helping users come up with creative ideas and solutions.

5. Google Gemini: Gemini can answer basic questions, help with coding, and solve complex mathematical equations.It also supports multimodal search, allowing users to input pictures as well as text into the chatbot.

Using AI has both positive and negative aspects that you should be aware of:

Pros

• Efficient Information Retrieval: can provide better results than search engines and can more readily point to web resources than a general search engine.
• Problem Solving: can provide analysis or a lot of data in much shorter amount of time that can lead to better problem solving and decision making.
• Personalized Learning: can allow students to engage in different ways and ask questions to help them better understand the material.
• Innovative Creativity: can act as a creative partner in developing new ideas, designs, and solutions.

Cons

• Hallucination: AI generated content is not always correct and can lead to errors or misinformation.
• Racial or gender bias: if the data used to train the AI is biased, the results it provides can also be biased.
• Shortcutting Learning and Academic Integrity: students can use AI to bypass the sometimes-difficult process of learning and may do so against AI policy set by a faculty member.
• Privacy: “free” systems often use your data as an income stream or to further train the AI.
• Job Displacement: AI systems can potentially replace humans in the workforce.
• Environmental Impact: AI systems consume a lot of electricity and is leading to an expansion of data centers and electrical demand.

• Responses are based on the information that was used to train the AI. This can lead to racial and gender biased responses based on the training material used.
• The sources used to generate responses are difficult or impossible to cite as the system does not provide a citation for what sources it uses to create responses.
• Intellectual property of others has been used to “train” the system without their permission.
• Since ChatGPT conversations are stored and used to train future models, any inaccurate information entered can later be reproduced as a response.

• Better web searching
• Creating summary of concepts for different audiences
• Summarizing articles and generation of key points and highlights
• Partnering in creation of new ideas and content
• Image and illustration generation
• Creating stories, poems, song lyrics, and images
• And more!

Since we subscribe to Microsoft 365 at UMW, everyone has access to Copilot, which has several advantages over other “free” GPT AI services.

• Copilot uses OpenAI GPT 4 – a more powerful version of the free ChatGPT 3.5
• Copilot has access to more recent information on the internet than ChatGPT 3.5
• Copilot also uses OpenAI’s DALL-E for image generation
• It provides a “protected” privacy mode – your data does not leak outside of UMW, chat data is not saved, Microsoft has no ability to see your prompts and data, and no data you enter into Copilot is used to train the model.

1. Go to https://copilot.microsoft.com
2. Login using your UMW email address and NetID.
3. Once you’re in, you should see your name and “Protected” in the upper right corner of the browser window.

Note: “Protected” gives your searches and data privacy protection. According to Microsoft “user and business data is protected and will not leak outside the organization. You can be confident that chat data is not saved, Microsoft has no eyes-on access to it, and it is not used to train the models.”

Here are some suggested prompts to start with.

Text Generation: “give me a summary of the Historic Preservation program at UMW”

Image Creation: “create an image of a cat wearing a virtual reality headset”

• Understand Co-Creation: Recognize that using AI is a co-creative process where both the user and the AI contribute to the outcome.
• Think Relationship, Not Tool: Approach interactions with AI as building a relationship, rather than just using a tool, to get more meaningful results.
• Experiment and Learn: Spend time experimenting with different chatbots and learn from the outcomes to improve your prompts.
• Seek Guidance: Since there’s no definitive manual from AI companies, seek insights from experts who have documented their experiences.
• Craft your Prompts: Add more detail to your prompts to further steer the AI to better responses. Vague prompts give vague answers.
  • Ask the AI to write for a specific audience.
  • Ask it to give you multiple versions of a reply.
  • Give it a persona or role to assume.
• Remember Your Role: You are responsible for what gets produced and shared, so do your due diligence to verify facts and citations provided by the AI.

Here are some items I used in creating this resource guide and presentation. I hope you might find them useful too.

How Should I Be Using A.I. Right Now? – The New York Times (nytimes.com) Ezra Klein Show podcast. 4/2/24.

Copilot in Bing: Our approach to Responsible AI – Microsoft Support – Microsoft. February 2024

One Useful Thing | Ethan Mollick | Substack

Wharton Interactive Crash Course: Practical AI for Instructors and Students – YouTube August, 2023

Copilot: AI Prompt Writing 101 | Microsoft Bing – Microsoft: 11/21/23