Introduction to Generative AI

Generative Artificial Intelligence (GenAI), often leveraging Large Language Models (LLMs) like ChatGPT, has emerged as a revolutionary tool with significant implications for various fields, including higher education. ChatGPT, specifically, gained widespread attention when it was activated in November 2022, disrupting the status quo in Higher Education as students began using it for academic submissions.

What is Generative AI?

GenAI models, or Large Language Models (LLMs), are essentially very large predictive models. They mimic human conversation by identifying language patterns and predicting contextual words. ChatGPT, for example, excels at generating coherent and relevant textual responses with minimal user input by leveraging an extensive database of trained language patterns. It generates statistically likely word sequences based on this database. It's crucial to understand that GenAI does not comprehend the meaning of text, nor does it think or reason; it simply generates text based on the relationships between words it has learned during its training.

Benefits and Promise of Generative AI

GenAI technologies hold immense promise for enhancing productivity and learning. In educational settings, they can offer numerous benefits:

• Personalized and adaptive learning experiences, improving student engagement, and reducing the burden on educators and administrators.
• Enhancing students' comprehension and fostering critical thinking skills.
• Automation and 24/7 availability, providing instant support to students and educators.
• Personalized assistance, helping students better understand complex topics.
• Scalability, allowing it to handle multiple inquiries simultaneously, making it an efficient support system.
• Cost-effectiveness compared to traditional educational support systems.
• Summarizing information in an easy-to-understand way, promoting self-directed learning.
• Speedy answers, saving time for both students and educators.
• Multilingual capability, ensuring accessibility for a broader range of users.
• Quick accessibility and availability, meaning support is always available whenever needed.

Limitations and Risks of Generative AI

While powerful, GenAI tools have significant limitations and risks that users must be aware of:

• Lack of Understanding: GenAI does not truly understand the meaning of text; it generates text based on statistical relationships between words. This limited contextual understanding can lead to incorrect or irrelevant answers.
• Not Sentient: Despite appearing to possess sentience or self-awareness, GenAI models are simply systems trained on large, potentially biased datasets. They are designed to output the most likely or common results, often suppressing less common or marginalized information.
• Bias: GenAI models carry implicit biases from their training data, making them unsuitable for ethical deliberation and decision-making. This data is also historical, which can result in a loss of context for current social changes. Their dependence on data quality and quantity can also lead to biases or limitations.
• Can Mislead/Hallucinate: GenAI often "hallucinates" or fabricates random data that is untrue, as models have no real sense of what is true or false. They are built to output what is most likely in a verbose manner, even if there isn't enough real information to support it. The lack of references provided by these tools makes it difficult to verify the accuracy of the information.
• Language Bias: LLM models are currently heavily biased toward Standard American English. This can penalize writing styles and dialects adopted by other cultures and ethnic groups, such as African American or Indigenous English, in favor of a privileged White-dominated form of writing.
• Privacy Risks: In most cases, the data you share with external GenAI tools is not private and can be accessible by external parties hosting them. Do not share private or sensitive information such as credit card details, ID numbers, or addresses. The model learns from its interactions, which may include sensitive information, and treats it as public domain data.
• Misleading Costs: Be wary of "free-to-start" tools or those requiring credit cards for free trials, as they may employ tactics to make subscription cancellation difficult.
• Not a Substitute for Critical Thinking: GenAI tools are just that – tools. They should assist your academic growth and should not replace your ability for critical thinking and problem-solving as an individual. It is essential to use them in conjunction with other educational resources and never to replace human reflection.

Ethical Considerations and Guidelines for Use

As GenAI is a powerful and potentially revolutionary tool, its ethical use is paramount. Here are key considerations and guidelines:

• Use Reputable Resources: Prioritize tools provided by your institution, such as U-M GPT, which are private, secure, accessible, equitable, and free for students. Data shared with these tools will not be used for model training and are not at risk of being leaked. When using external tools, ensure they are legitimate and follow safe computing practices.
• Clarify Purpose: The use of GenAI should align with teaching and learning goals, enhancing outcomes and providing support, not replacing human interaction.
• Emphasize Critical Thinking: Encourage students to evaluate information from GenAI and develop their own ideas and perspectives.
• Transparency and Guidance: Be transparent about how GenAI works, what data is collected, and how it will be used. Students should be guided on effective use, interpretation of results, and how to avoid pitfalls. Your professor should communicate rules regarding GenAI use in the course syllabus.
• Verify Information: Do not cite information from GenAI as truth. Always check any citations it provides and conduct your own research. Librarians can also assist. Cross-check information with other credible sources and use GenAI as a supplement, not your only source.
• Respect Privacy: Ensure that the use of GenAI respects student privacy. Any collected data should be stored securely and used only for its intended purpose.
• Open Discussion of Ethics: Discuss the ethical implications of GenAI openly, including its limitations, potential biases, copyright concerns, labor implications, environmental impacts, and data rights. Critically evaluate the quality of results and adhere to academic integrity policies.
• Personal Reflection: Ask yourself:
  • Is using GenAI helping me learn more and think better?
  • Is it enabling or hindering my mastery of course objectives?
  • Is the content generated accurate, verifiable, and free of biases that might harm others?
  • How will I treat content that might have been generated using GenAI?
  • Is its use equitable to my peers?
  • How can my actions in using GenAI lead to the greater good of society?

Understanding and adhering to these principles ensures that Generative AI serves as a valuable tool for learning and development, rather than a hindrance or a source of misinformation.