1. Seattle Colleges home page
  2. Employees
  3. Faculty eLearning & Instructional Design
  4. AI Task Force
  5. AI Information For Students

AI Information For Students

banner Seattle Colleges AI Task Force, skyline with space needle, airplane, and robot

Adapted from North Seattle College's Library Guide on AI by Peter Oliver

What you need to know

Generative AI is increasingly embedded in common tools such as Microsoft 365, Grammarly, and Spotify. Newer tools like ChatGPT, Gemini, and DALL-E2 are powerful, with an impressive ability to produce content based on prompts.

However, using generative AI tools (especially LLMs such as ChatGPT) in your academic work requires careful consideration. 

Before using AI for any coursework

Have a conversation with your instructor about AI use. If you are unsure whether AI tools in general are allowed in your course, or if you want to use a specific AI tool that isn't mentioned explicitly in the syllabus, reach out to your professor. Having conversations early is the best way to avoid confusion and problems down the road.

The more you know

Most AI tools do not have access to information behind paywalls, including much academic literature: Because of this, a rigorous search for sources requires the use of academic databases and search engines.

Tip 💡

Explore AI software and tools to understand what they can and cannot do. Use a topic you already know a lot about. Take the time to critically analyze their responses. Ask the same question of the same AI tool in different sessions and compare the answers. Ask follow-up questions.

Three things to keep in mind about AI tools

  1. They have uses, but also limitations

  2. They are fast, but can be wrong and have biases

  3. They are easy to use, but can outsource your learning

Practical Intro to AI

You can watch this 10-minute video to gain a basic understanding of how AI works and how it is impacting teaching, learning, and work. 

Citing AI

You should cite AI tools whenever you use them to:

  • Gather or present information

  • Write text

  • Edit Text

  • Synthesize ideas or find connections

  • Clean or manipulate data

Here's a more in-depth guide to citing AI

Some ways to explore Ai chatbots

You can think of generative AI chatbots like ChatGPT as an assistant or brainstorming tool. It doesn't really know the answers, but can help you solve problems. 

Here are example prompts you can try:

  • Give me 5 titles for a college-level paper on [insert subject and topic]

  • Organize these notes into an outline: [paste your text] 

  • Create flashcards from these notes: [input your notes]

  • Explain the difference between [concept X] and [concept Y].

  • Write a mnemonic device to help me remember the difference between these two concepts.

Here are example prompts:

  • Give me five writing prompts about [topic]

  • Provide practice problems for [subject]

  • Check my description of [topic or concept]

To write better prompts, get more specific. Add a quantity to your request, the genre of your request, or format:

  • Give me five creative writing prompts about [topic]

  • Provide ten multiple choice practice problems for [subject]. Make some easy and some intermediate.

  • Check my description of [topic or concept] for accuracy and make a table of any errors and possible improvements.

Here are example prompts:

  • Provide some common criticisms of [specific concept or issue]

  • Provide clear, multiple paragraph explanations of [concept you what explained] using specific examples. Then, give me five analogies I can use to understand the concept in different ways.

Caution ⚠️

Most AI tools have data only through a certain date. If you ask a chatbot about a recent event or subject outside its input window, it will make a guess that will likely be inaccurate. But the chatbot will sound very confident and convincing! It's always a good idea to ask the tool: Are you sure?

Caution ⚠️

Providing an AI tool with your own original work can pose privacy concerns. Your inputs are stored and may be used to develop future versions of the tool. 

Never put anyone's name, address, phone number, and other personal information into an AI tool! You can find out more by visiting this guide on Data Privacy for students.

Can I use ChatGPT instead of Google?

AI chatbots are marketed as question-and-answer tools, but they are not search engines (even though they are sometimes returned with Google search results). Here is an overview of how they compare:

Search engines (Google, Bing, etc.)

AI chatbots (ChatGPT, Perplexity, etc.)

  • "Trains on" text from the internet and other large sets of data, then calculates statistical probability from patterns found in the data ​

  • Designed to model information and then generate likely outputs based on the model​

  • Makes it difficult to reliably use the information in the text that is AI-generated ​

  • Requests for sources, like articles about a topic, are treated as a task of generating the most probable text - hence "hallucinated" sources that sound right but don't exist

Common biases in AI

There are several organizations that tackle equity issues in the global AI industry, such as the DAIR Institute and Black in AI.  Here are some of the persistent issues with AI tools:

  • Sample Bias: The input data (text, pictures, etc.) that is used to build and train the AI tool only provides an incomplete sample of the world, even if a very large dataset was used.

  • Programmatic Morality Bias: This bias is added by the creators of the AI tool, sometimes to try to correct for sample bias. There can be over- or under-correction of bias, which leads to unintended consequences in the text or images that the tool produces. ​

  • Ignorance Bias: If something isn't explicit in the input, it won't be in the output. AI is not capable of deductive reasoning.​

  • Overton Window bias: Output stays within the confines of what is “possible” according to biases 1 through 3.

  • Deference bias: Users tend to overly trust and value AI contributions, in spite of the biases listed above.

Source

You can learn what algorithmic bias is in this video:

Uses and limitations of AI tools

Uses

  1. A starting place for brainstorming ideas and outlining a project

  2. Simple text editing, debate role-play, and tutoring

  3. Writing or debugging code

  4. Entertainment

Limitations

  1. Not a search engine and not a substitute for research:

    • Can give you false information ("hallucinations"), which are difficult to notice if you aren't already an expert on the subject.

    • Can give you convincing-looking citations that are fake (best to use them as suggestions to look up and explore yourself).

  2. Can interfere with your learning by causing you to skip steps.

    • Must be prompted to give you feedback in addition to corrections.

    • Must ask for very specific feedback to get good tutoring.

  3. Can be inconsistent, make errors, and fail to teach you how to fix problems.

  4. Many tools are made by for-profit companies. 

    • Privacy policies can allow AI companies to collect and profit off your personal information. 

    • The environmental costs of AI are significant. Learn more about the ethical aspects of AI

    • By engaging with them and adding to their training data you give companies unpaid labor.