Want to get started using GenAI, or just learn more about how it works and what it might mean for higher education? This is the place to start! On this page, you will find an overview of the technology, including information about what it can and can’t do. You will also find some suggestions for how you can take some first steps in putting it to use through experimentation, effective prompting, and keeping an eye out for some of the potential issues with GenAI outputs.
What is GenAI?
GenAI is a type of machine learning that generates new content by analyzing and processing vast amounts of data from diverse sources. GenAI tools can generate text, images, video, sound, and code. Different tools are trained on different datasets and with different training methods. These tools work by identifying the patterns and structures in their training datasets and using this learning to generate new outputs with similar characteristics. Because the outputs of these tools are probabilistic – they are guesses about what comes next based on the patterns the tool has seen in a large data set – they may contain factual errors and biases. The quality and diversity of the training data significantly impact the performance and reliability of GenAI tools.
Large language models (LLMs), such as GPTs, specialize in analyzing and processing text and generating new text. OpenAI’s popular ChatGPT and Microsoft’s Copilot are chatbots that allows users to interact through a chat interface with the LLM. There are also multimodal machine learning models that integrate multiple communicative modalities, including linguistic, acoustic, and visual messages. For example, tools like Midjourney can create visual outputs in response to text-based prompts.
York University staff and students have access to Microsoft Copilot with enhanced data protection, via your Passport York ID. Copilot uses the latest GPT model and is available for free.
This video provides a helpful introduction to GenAI for educators and students.
What it can do
GenAI has a wide and evolving range of capabilities. While different tools have specific strengths and foci, typical applications in postsecondary education may include:
- Acting as a personalized tutor or brainstorming partner
- Summarizing course readings
- Creating or debugging code
- Outlining and planning projects
- Drafting, co-writing, or polishing written text
- Drafting lecture materials, test questions, and study aids
- Tailoring course content to student learning needs
- Generating rubrics
- Assisting in providing feedback
- Analyzing datasets
- Facilitating literature reviews
What it can’t do
GenAI cannot replace the nuanced understanding and empathy of human educators. It struggles with grasping context-specific subtleties, cultural sensitivities, and the emotional aspects of teaching and mentoring.
Some drawbacks to be aware of as you explore GenAI include:
- Because it is probabilistic, GenAI is prone to “hallucinations” and can make false yet very probable sounding claims with confidence
- AI cannot independently verify the accuracy or relevance of the information it generates, requiring human oversight
- GenAI may reproduce the biases inherent in its training data and can create content that contains both obvious stereotypes and more subtle implicit or systemic biases
Making use of Gen AI
While GenAI is not new, OpenAI’s launch of ChatGPT in November 2022 marked the fastest recorded adoption of a technology tool to date. Since that time, the following GenAI tools have emerged as some of the most dominant. Of these tools, only Microsoft Copilot is currently endorsed by UIT for use by the York Community. Visit UIT’s Artificial Intelligence page to learn more about Microsoft Copilot and other AI tools. Try some hands on experimentation to get a sense of how the tools differ!
Effective prompting
Getting the most out of GenAI tools is not just about what you ask for, but also how you ask. Crafting effective prompts lead to more accurate, relevant, and useful GenAI outputs. Here are two basic approaches to prompt engineering. For more ideas, see the Learn More section, below.
- Conversational Prompting: You can get great results just talking to the AI chatbot about what you want and going back and forth to refine the results. Two tips for effective conversational prompts:
- Provide some context for your request.
- Ask the GenAI to ask you for the information it needs – end your prompt with something like “Before you begin, ask me any questions that would help you do this better. Be very thorough.”
- Structured Prompting: These prompts are carefully crafted, with multiple elements, to help the GenAI succeed at a specific task. Several models for structured prompts have been developed. Common elements include identifying the role the AI is to take on and identifying the context and requirements of the task. Check out the resources below for guides on structured and conversational prompting.
Evaluating outputs
GenAI tools require human oversight and accountability, and assessing their outputs is an important aspect of this. GenAI can present flawed outputs with great confidence, and all users should consider:
- Sources and citations: Are the references provided by GenAI real, accurate and appropriate?
- Content accuracy: Can the claims made by GenAI be corroborated by a trusted source?
- Relevance and coherence: Does the output meaningfully engage with the task assigned?
- Bias: Does the GenAI output explicitly or implicitly reinforce any stereotypes or biases?
Want to Learn More?
Prompt Library. More Useful Things: AI Resources
A range of pre-formulated prompts for using GenAI as an aid with class preparation and teaching, student exercises, and other tasks, from Ethan Mollick’s blog.
York UIT’s prompt library is a free resource available to anyone. Prompts are searchable and organized by tags. You can even submit your own prompts to grow the collection.
Empowering Educators by Harnessing Generative AI Tools: Navigating Prompt Roles
From the National Center for AI, a how-to guide for educators getting started with prompting GenAI, with description and examples of conversational prompts and structured prompts (also called prompt roles).
What is ChatGPT Doing… and Why Does It Work?
For those seeking a thorough explanation of how GenAI works, an in-depth explanation from the creator of Wolfram Alpha.