Skip to Main Content

Teaching and Generative AI

Resources to help SJSU Faculty and Teaching Assistants how to address ChatGPT and other generative AI tools in their classrooms and assignments.

Welcome to the Teaching and Generative AI LibGuide!

ChatGPT Mobile Screen

This guide is designed to help you:

  • Evaluate whether and how to incorporate tools like ChatGPT into your courses.
  • Make informed decisions based on your specific pedagogical goals.
  • Design assignments that thoughtfully integrate generative AI tools.
  • Address concerns around inappropriate or unauthorized use of AI in academic settings.

 

As of May 2025, SJSU does not have a specific policy regarding student use of AI. California's public higher education systems have not yet created a formal policy regarding using ChatGPT and other generative AI tools. Many universities leave the decision to apply generative AI tools in assignments and courses up to individual faculty and teaching staff. However, it is highly encouraged for faculty to clarify their stance on generative AI and other technologies in their syllabi.

What is AI? What is Generative AI (GenAI)?

AI is the ability of computer systems or algorithms to perform tasks that typically require human intelligence, such as learning, reasoning, problem-solving, and decision-making. For example, reading handwriting, recognizing speech, and providing Netflix recommendations. Artificial Intelligence (AI) is everywhere and it's actually has been around for a long time.

Between the 1950s and 1990s, the field of artificial intelligence was dominated by what is now referred to as classical or symbolic AI. These systems operated using explicit, human-programmed rules and logic to perform tasks. This rule-based approach enabled early AI to automate specific functions, play games, simulate conversation, and respond to queries. In the decades since, AI has advanced significantly, giving rise to modern techniques such as machine learning, natural language processing, computer vision, and generative AI(GenAI).

Generative AI is a form of artificial intelligence that creates new content such as text, images, music, video, code, and voice. It responds to user input, often called a prompt, and can produce content in different tones, styles, and formats. These tools are capable of synthesizing information, answering questions, and adapting to the way a prompt is phrased. Examples include ChatGPT for text, DALL·E for images, and Suno for music.

Unlike earlier AI systems that mainly analyze data or make predictions, generative AI produces original outputs by recognizing and reusing patterns it has learned from large collections of data. Although its responses can appear creative, novel, and "intelligent," it is often argued that the content is not entirely new but rather a recombination of existing material based on training data.

Generative AI models are used in a wide range of applications, from writing assistance and research support to visual design and media creation. These models have also significantly improved in recent years. Originally, large language models (LLMs) like GPT-3.5 and earlier versions were limited to text input and output. Newer models, such as GPT-4 with vision, Google Gemini, and Anthropic Claude, are now designed to handle multiple types of input, such as text and images, and in some cases can generate non-text outputs like images, charts, audio, games, or apps. This emerging capability is known as multimodal AI, which refers to models that can process and respond to more than one type of input and not just the ability to generate text.

Below are some of the major types of generative AI models and tools in use today.

Disclaimer: This list is intended for informational and educational purposes. SJSU and the King Library do not endorse any specific generative AI tool or provider mentioned above.

Text Generation (Large Language Models): Models are designed to understand and generate text. They are commonly used for writing, summarizing, tutoring, coding, and research.For example, GPT-4o/GPT series (OpenAI): Powers the ChatGPT application, which is capable of advanced reasoning, writing, and conversational tasks. GPT-4o also includes vision and voice capabilities, making it closer to a multi-modal language model. Other examples are Gemini (Google) suite of LLMs, available in both text-only and multimodal versions.

Image Generation: Models generate visual content from text prompts. Examples: DALL·E 3 (OpenAI):High-quality image generation model integrated into ChatGPT, Stable Diffusion (Stability AI):Open-source text-to-image model, Midjourney, and Adobe Firefly.

Adobe Firefly generative model is focused on commercially-safe content creation within design software. Available with SJSU subscription.

Music Generation: Converts text prompts into full songs, including vocals and instrumentation. Ex: Suno AI, Udio Video Generation: Models generate or edit video content based on text, images, or a combination of inputs. Examples: Runway Gen-2, Pika Labs

What can Generative AI do?

  • Answer your questions
  • Create content (poems, essays, freestyle raps, stories)
  • Help you proofread and rewrite paragraphs (emails, cover letters)
  • Assist with brainstorming
  • Suggests computer code to perform a task
  • Explain and tutor
  • Create an app or game

What are the limitations? 

  • Lack of context: AI does not understand meaning or emotion the way humans do.
  • Bias: Generated content may reflect biases in the training data.
  • Hallucinations: AI may generate content that sounds accurate but is completely false or misleading.
  • Makes up citations: AI may fabricate citations or research works that do not exist.
  • Limited access to verified research: Many AI tools do not have access to scholarly databases or subscription-based academic resources.
  • Dependence on training data: AI cannot go beyond what it has learned.
  • Environmental impact: High energy and water consumption in training and running large models.