AI in Teaching: Learn About AI

Appropriate use of generative AI begins with understanding how it works. AI technologies that generate text, images, and other creative works use a process known as machine learning to review a vast range of content and identify patterns and structures, accounting for nuance and context. They respond to natural language prompts by using statistical inference to predict and simulate human-like dialog or imagery. Their outputs are based on probability and do not actually entail conscious creativity or understanding; therefore, they always come with a statistical probability of being inaccurate, biased, or simply off the mark. However, the accuracy of generative AI tools tends to improve as new AI models emerge. People who use AI effectively understand its limitations, but they apply critical thinking and domain knowledge to evaluate its outputs and can prompt it to produce useful results. They also accept accountability for the content they produce with AI assistance.

Getting Familiar with AI Tools

Some of the leading AI chatbots have been integrating a mix of tools that can produce a variety of content and responses. Exposure to, and familiarity with these tools is an important starting point in becoming familiar with generative AI:

Beyond these and other major chatbots, specialty AI tools exist for image generation, coding, audio, video, email management, and much, much more. Here we will list just a small handful of image generators, coding tools, and commercial software tools now incorporating AI that students or professionals may access and use.

AI in commercial software

The tools listed above are standalone generative AI tools. However, generative AI capabilities will increasingly appear within everyday software tools and services. Current examples include:

  • Copilot in Microsoft Office 365 (available by request and with a monthly fee through USU)
  • AI in Box, available for beta test by request
  • Generative fill in Photoshop and After Effects; Generative Color in Adobe Illustrator
  • Generative AI in Grammarly
  • Generative images in Canva and Adobe Express

Generative AI will likely appear in most widely used software applications and services in the near future.

Image Generators

Coding

For more resources see a list of USU’s supported AI tools.

Prompt Engineering, Iteration, and Review

The way you write your prompts makes a difference in the output you receive from a generative AI tool. Prompt Engineering is the practice of crafting effective prompts to get more desirable outputs.

Tips for Prompting AI Chatbots

  • Be Specific: Clearly define what you need. For example, “Explain the concept of photosynthesis at a college freshman level” is more effective than “Tell me about photosynthesis.”
  • Context Matters: Provide necessary background information. If you’re discussing a specific theory or event, include relevant details in your prompt.
  • Guiding the Output: If you want the response in a specific format (e.g., bullet points, a short essay), state this in your prompt.
  • Use Follow-Up Questions: Engage in a dialog with the chatbot. Follow-up questions can refine the response or explore different aspects of a topic.

Some chatbots also allow the use of parameters to influence the final output. For example, learn about ChatGPT parameters, such as temperature, max tokens, Top P, frequency penalty, and presence penalty. Also see Google's getting started guide for prompting in Gemini.

OpenAI provides an excellent guide on prompt engineering, which includes numerous additional tips, such as asking the the model to assume a specific persona, providing examples, asking the model to use a reference text, splitting tasks into simpler subtasks, and giving the model time to "think" by prompting it to engage in an inner monologue or work out solutions before rushing to a conclusion.

Tips for prompting AI Image Generators

  • Detailed Descriptions: The more detailed your description, the more accurate the image. For instance, “a tranquil beach scene with a sunset, palm trees, and a hammock” gives the AI image generator more clarity about the picture to create.
  • Style and Aesthetics: Specify if you want the image in a particular artistic style (e.g., impressionist, realistic).
  • Iterative Process: Your first prompt might not yield the perfect result. Refine your descriptions based on the initial output.
  • Ethical Considerations: Avoid prompts that could lead to the creation of inappropriate or harmful content.

AI Image generators also allow the use of specific parameters to achieve specific results. For example, Midjourney has an extensive parameters list.

The Importance of Review and Iteration

As noted above, generative AI does not use magic—it uses probability. It also works within bounds, generating only so many tokens of content at a time. Its outputs are often incomplete, and they can also be incorrect or fabricated. Effective use of AI requires domain knowledge in the area the AI is being prompted to assist with. Do not automatically trust AI outputs, but check them for accuracy and omissions. Often the first prompt will not achieve the most desired output. Review the output of the prompt, ask for details and clarifications, provide examples, and break large tasks into smaller tasks. Sometimes you can return to an earlier prompt after some dialog and achieve a better result.


Privacy, Security, and Ethics

The potential applications of generative AI are many and powerful. But with great power comes great responsibility. All users of AI, including teachers and students, should understand the implications and nuances of privacy, security, and ethics when using these tools.

Care for Personal Data

Generative AI models may retain or recall information from previous interactions, and can inadvertently expose personal data present in their training datasets. Avoid inputting any information that could reveal a person's identity or personal details, including your own. Also, ensure that any text or image generated does not include sensitive information.

As a best practice, use hypothetical or anonymized data for exercises, demonstrations, research, and content development.

Informed Consent, Copyright

Be sure to read the terms of service and privacy policies of the tools you use or ask your students to use. Some tools, such as open-community image generators like Midjourney, make their outputs publicly viewable and allow other users in the community to reuse and remix those outputs. They also allow individuals to file copyright complaints takedown requests.

Users of AI should obtain informed consent from individuals before using their image, creations, or personal information to generate AI content. They should also avoid creating and publishing content that would infringe upon the copyright of another content creator. 

Accountability for AI Content

Users of generative AI are accountable for the content they produce and publish with its assistance. They can be held legally liable for their use of AI. If a user republishes false, defamatory, outdated, plagiarized, copyright-infringing, biased, or otherwise harmful content produced by AI, the user is generally considered responsible.

Therefore, AI users should verify the accuracy of the information provided by AI. They should review it for deceptive, harmful, or biased content. They should also be aware of the policies and regulations of the venue in which they are working regarding content originality, plagiarism, attribution, and so on.

Avoiding Malicious Use

AI can be used to create deceptive content such as deepfakes, falsified texts, falsified research, misrepresentations of one's own capabilities, and more. Such activities are unethical, often against the terms of use for generative AI platforms, and sometimes illegal. Students should be discouraged from using AI to these ends.

Remaining Informed About AI

Generative AI can be a powerful ally in education when understood and used responsibly. AI technologies are evolving rapidly, and the issues, regulations, and best practices surrounding them are also in a state of flux. Wherever possible, educators, students, and professionals should participate in community forums and follow industry happenings relative to AI in their respective fields. Regularly using AI technologies is another great way to remain informed about them.