Two recent studies at the University of Maryland and Cornell University analyzed the efficacy and rate of false positives with all the prevalent AI detectors for writing on the market today. This strengthens the argument that we as teachers need to know our students and their abilities to make the call or start the conversation with them when we doubt the legitimacy of their work. Even with the best AI detector on the market, there is a more than 20% chance that you will wrongly blame a student for cheating. If you are wrong you run the risk of damaging your relationship with that student forever which can negatively affect their performance at school and beyond. We don't want to trust the basis of our relationship with our students because of an app that doesn't give us complete certanty.
If there is doubt, this should be the start of the conversation. Right now the Google extension DraftBack is one of the best options to truly see how much work a student has done on a document. Here is a tutorial video that shows the ways that you can use this tool to figure out if a student has been honest or not.
Generative AI Models
Most Generative AI tools are derived or plugged into one or more of these models.
Artificial Intelligence has become a massive buzzword in society since Chat GPT was launched to the public in November '22. Artificial Intelligence (AI), like ChatGPT, Gemini, Llama, Microsoft Copilot, and Claud are transforming education by enabling personalized and interactive learning experiences. AI tools can automate administrative tasks, offer customized feedback, and accommodate various learning styles, making education more accessible. These larger AI tools are becoming embedded in programs we already use like Canva, Grammarly, and even possibly Google Workspace. Some of you may see it embedded in your personal Google environment. For middle school teachers, integrating AI can enhance teaching methods and student engagement. This webpage introduces a selection of user-friendly AI tools designed to complement traditional teaching and enrich both instruction and learning. Embrace these innovations to explore new educational possibilities in your classroom while giving student the opportunity to explore this tool that they will most definitely be using in the years following school.
Generating effective prompts to get what you really want out of a GPT
Generating an effective prompt is not as easy as it sounds and takes some practice. Very simple prompts will give you responses with a lot of variability and will increase your chances of getting an answer with AI Hallucinations. Here is a guide from OpenAI that gives guidance of the parts you should include to generate a specific prompt that will give you the most effective response from a GPT. This is another great article that goes into more detail about the "why" regarding detailed prompts.
This article is a great starting point. 3 Helpful Teacher AI Prompts From Short Answer
Prompt Engineering Framework
This slide deck simplifies the information from Dan Fitzpatrick who uses the PREPARE acronym for creating detailed prompts in Generative AI.
It outlines seven domains needed for a solid prompt; Prepare, Role, Explicit, Parameters, Ask, Rate, and Emotions. Then it offers an activity that you can do with students or other staff members to see the differences in the output you get as you keep adding more and more detail to the prompt. This could be done with any of the generative AI models listed above.
Differentiate some assignments that allow the use of AI and others that don't at all.
The Red - No AI
Best for assessing foundational knowledge and skills
Not necessarily for teaching foundational knowledge and skills
Using GEN-AI would reduce the validity of the assessment
The assessment portion of the assignment should likely be completed in class OR with AI guardrails in place
Weight?***
Most work going in grade book should be red
The Yellow - Some AI
Best for assessing tasks that must be completed over time (i.e. long-form writing)
Requires clear guidance and dialogue on the expected use
A reflective element on AI should likely be part of the assessment if going in the grade book
Weight? *****
Some work going in the grade book should be yellow
The Green - Full AI
Best for tasks whose primary goal is in teaching students to use AI effectively
Requires clear guidance and dialogue on the expected use
A reflective element on AI should likely be part of the assessment if going in grade book
Weight? *****
A small percentage of work going in grade book should be green (unless it’s digital tools class)
Concepts and graphics borrowed from Adam Sparks's presentation "Writing Instruction for the Age of AI"
Experts vs. Novices in the World of AI
This chart illustrates how a novice using AI, without a good foundation of fundamental knowledge, gets a false sense of expertise. This is like our students using AI to write papers for us. They think they have a great product but without practice of writing conventions like tone, voice, and content knowledge, use of AI keeps them as a novice and shunts their growth because they can't effectively critique the output from AI or they can't generate better prompts that will produce a truly accurate and effective response.
Experts on the other hand that have a good basis of fundamental knowledge of the content they are working with, become super experts when using generative AI. They can effectively critique and expand upon the output and can then generate more specific and detailed prompts to elicit detailed prompts to get tailored responses from AI.
By giving our students the opportunity to use generative AI in a guided framework, with a teacher, we can help them see that until they truly know how to create good writing and answers on their own, using AI will actually make it harder for them to be an expert at anything.
Academic Integrity and AI
In the era of artificial intelligence, discussing academic integrity in schools is more crucial than ever. With AI tools capable of generating content and solving complex problems, it's essential to educate students about ethical use and the value of original work. Emphasizing academic integrity helps students develop critical thinking skills, maintain trust in educational environments, and prepare for a future where ethical technology use is paramount.
This article introduces the latest version of the AI Assessment Scale (AIAS), a framework that helps educators integrate artificial intelligence into academic assessments. Updated to reflect advancements in AI, the scale provides five structured levels, guiding institutions on using AI ethically and effectively in both K-12 and higher education contexts.