Getting to Know GenAI

Generative AI can do far more than produce middling essays or solve homework problems.

Developing a clear understanding of GenAI's capabilities is essential for making informed decisions about how to address its use in academic, instructional, or organizational contexts.

Recent findings from the SDSU AI Student Survey indicate that more than half of students are already using GenAI tools in their academic work. Many describe using GenAI as writing support. Grammarly, for example, was cited by 37% of respondents as a tool used for academic work, nearly as often as ChatGPT.

Others use GenAI as an on-demand study partner. One respondent noted, "If I’m struggling with a question, I typically ask AI, which gives me research on the spot." Participants also highlighted how these tools provide clear examples, break down complex concepts, and support problem-solving in areas such as mathematics and coding.

Taken together, these uses point to the remarkable versatility of GenAI—and that breadth can feel overwhelming at first. Before we explore specific applications in more detail, it’s helpful to have direct experience.

If your engagement with chatbots has been limited so far, or if this is your first sustained exposure to them, now is a good moment to experiment. Open a chatbot, try a range of prompts, and observe how different tools respond. This hands-on familiarity will provide a shared foundation for the work ahead.

Step One: Explore

Chatbots

For this exercise, there are a number of chatbots that you may consider exploring, including:

If you haven’t already done so, take a few minutes to play. Just give your chatbot an instruction.

The instructions you give are called prompts.

Each prompt contains the information or input you give to a chatbot (or image or code generator) to shape its response or output.

You can think of prompts as directions for a personal assistant: a well-meaning agent who needs specific guidance and whose work may sometimes contain mistakes, no matter how carefully we instruct it.

The user's domain expertise (prior knowledge of the topic they are working on with GenAI) can be valuable for setting the right parameters for a prompt. Domain expertise is particularly helpful in evaluating output.

In terms of prompt phrasing, it’s always nice to say please!

Plus, as summarized by Fox Cahn, evidence that technology shapes cognition suggests that if we become too accustomed to "cold precision," we will end up "treating our fellow human beings with the [same] apathy, disrespect, and incivility" afforded to our GenAI conversation partners.

This scenario aside, and however you prompt, don’t ask for simple facts: remember, chatbots aren’t search engines or simple web browsers. They don't just retrieve; they generate. They augment our productivity or performance.

If you need some ideas: ask your chatbot to generate a short song about a topic you love, written in the style of a songwriter you admire. Or you can ask the chatbot to take a discussion post or document you've written and make it more concise, more (or less) formal, and comprehensible by a kindergarten student, or turn it into a rhyming poem.

In other words, take a few minutes to goof around.

Step Two: Reflect

Once you've had a chance to experiment, reflect on any divergences between what you expected and what you experienced.

Did rewording or augmenting the instruction (with specifics regarding length, tone, and format) lead to the output you were looking for?

If you haven’t given a follow-up instruction, please do so now (e.g., "Please make the song half as long," "Show the information in a table," or "Explain [x] again, but this time to a 5th grader").

Don’t spend much time on this—try it once or twice—because we will soon explore GenAI in a more structured way.

General Rules

While playing just now, you will have observed that the more specific your instructions were, the more likely the chatbot was to generate output that met expectations. Let's think about that.

In doing so, consider these two key principles regarding input and output:

Be specific in your prompts:
- The chatbot does what it’s told; so specific parameters are helpful (e.g., ‘generate a table’, ‘I’d like a bulleted list’, ‘provide 10 true/false questions’). Further, the app uses what it’s been told before to shape later responses; its output is iterative in terms of what you’ve already provided (e.g., ‘Which five of the 10 questions are best for [fill in the blank]?’).
Use GenAI responsibly:
- Verification involves more than just checking for face validity or whether the output feels right. Generative AI can put things together in logical but meaningless ways.
- The term hallucination refers to a GenAI response that, while grammatically correct, is counterfactual. Remaining vigilant for such counter-facts is a crucial part of responsible use.

Click to Learn More: The "EVERY" Heuristic