Navigating AI

Most artificial intelligence systems operate via command. The user will often make an account and then be able to ask the system to perform a certain task. In cases, such as Microsoft office, the spell/grammar check AI will make note of possible errors within a text and provide suggestions for the user.

The AI will usually outline instructions and highlight its capabilities/limitations for the user. Additionally, the AI may improve its system via user feedback.

The best results from an AI are based on the nature of its commands. A user must be as specific as possible with an AI to yield as close to their desired results as they can get. 

Below are some things to keep in mind when it comes to using AI:

• Access AI Website 
  • E.g. Google "ChatGPT" or visit chat.openai.com
• Login/Signup: Use email, Google, or Microsoft account
  • ChatGPT Chat Area: 
    • Examples, capabilities, limitations, ask questions, provide feedback
• Try Again: ”regenerate response” or enter refined prompt

Artificial Intelligence systems “synthesize” text and use patterns to formulate their responses or actions (Schroer, 2023). AI written content tends to be more formal and repetitive than human writing. Additionally, the AI system tends to lack originality or context because it is sorting through information that already exists.

Many companies have used artificial intelligence to detect AI written content, but these systems have been subject to “false positives” and other limitations such as requiring a word minimum, inaccuracy with human-edited AI text, and capable of only surveying one language style (typically English). In addition, AI systems can store information submitted to them and use that information for other purposes, such as training their AI models.  Submitting student information to an AI system could constitute a violation of FERPA unless the system is approved for such use through university agreements.

The best methods and approaches for detecting academic plagiarism based on our research have been experimenting with the AI ourselves and refining how our assessments are created /administered. For example, faculty in composition programs often have students submit ideas, outlined, and multiple drafts before their final paper.  This process demonstrates students' thought processes and provides additional opportunities for feedback.

Below are some things to keep in mind when it comes to detecting AI:

• AI Writing: uses patterns seen in other texts
  • Formal, generic, repetitive, tends to lack context
• Best Methods: 
  • Trying AI out with desired prompts
  • Refining how assessments are created & administered
  • Focusing on the process of constructing knowledge instead of the final product
• Refer to Guidance for AI in Teaching, Learning, and Scholarship

Artificial Intelligence does have some pitfalls that developers are hoping they can work out in the years to come. For the most part, each system lacks originality because it is sorting through existing information. This also lends itself to providing incorrect information as it takes the existing text to fill in its gaps in knowledge.

Some users may find AI responses “harmful” or “biased. Many AI systems can only regurgitate information within a certain timeframe because its “data” was set to operate within that time period (e.g. ChatGPT does not know about events or trends past 2021) (OpenAI, 2023). 

Lastly, artificial intelligence tools, such as ChatGPT, do not specify what reference sources they are using so the user may need to verify this information outside of the tool and cite those sources.

Below are some things to keep in mind when it comes to the limits of AI:

• May lack knowledge past a certain timeframe 
  • E.g. ChatGPT doesn't know of events or trends past 2021
• May lack originality
• May not be able to evaluate itself
• Sensitive to input, may require specific prompts
• Inaccurate information
• Doesn’t specify what sources 
• May provide biased responses 

Given all this information, about artificial intelligence, we may need to refine the way we create and evaluate our educational assessments. Bloom’s Taxonomy is an excellent resource to use when thinking about assessment creation. Original work tends to favor more creation, evaluation, and analysis work.  For more information regarding Bloom’s Taxonomy, check out the following site:

Bloom's Taxonomy Revised

Another item to keep in mind regarding assessments is how students are being evaluated. Perhaps we can move to assess our students on their active participation or process. Some examples include evaluating live videos at a specified location, instructing to interview a specific person or organization, prompting a student to deliver a presentation, or facilitating based on the nature of the student’s knowledge or series of questions versus the presentation's aesthetics.

AI technology can also be used by the students to think critically by having them analyze an AI’s response on a given topic or having them figure out how the AI came to this response. Students can also use AI as a tool to complete an overall project like how they use calculators or spell check. Tools, such as ChatGPT, are great ways for students to mind map and outline their ideas so certain assignments could ask for a timestamp of this process with the tool.

Below are some things to keep in mind when it comes to refining assessments:

• Assessing on active participation & process: 
  • Examples: Live videos, interviews, presentations, facilitations
• Critical Thinking: 
  • Having to use specific readings or sources 
  • Have students analyze/evaluate AI-generated responses 
  • Find out how AI came to a response
• Using AI as a tool: 
  • Completing assignments using AI (e.g. time stamp, mind mapping, outlining etc.)

Improvements are constantly being made by AI tools and programs. Some tools, like ChatGPT have already made updates to better their ability to reason, concise information, and exhibit creativity. AI is not going anywhere, and it is for this reason that we should continue to become familiar with artificial intelligence while reflecting on the ways we can use these tools for more opportunities at work and in education.

Below are some things to keep in mind when it comes to the future of AI:

• GPT-4: 
  • Improved reasoning/creativity, concise responses 
• Ongoing Journey: 
  • Continuous improvements to AI
  • New opportunities for growth in work/education

 Our team is performing on-going research on artificial intelligence and examining its impact on higher education.

Recording of how to use ChatGPT-4 to create a PowerPoint as of March 2023.

OpenAI . OpenAI. (2023). Retrieved March 22, 2023, from https://openai.com/

Schroer, A. (2023, March 3). Artificial Intelligence. BuiltIn. Retrieved March 22, 2023, from https://builtin.com/artificial-intelligence