Lesson 2: Interacting with Language Models

Getting Started with Code.org: Consider watching our Getting Started with Code.org video series for an overview of how to navigate lesson plans, setup a classroom section, and other important features of the Code.org platform. Each video also has a support article if you'd prefer to read or print instructions - click here to learn more.

Setup a Classroom Section: You can use a class section in Code.org to manage your students, view their progress, and assign specific curriculum - click here to learn more.

If you are using a learning management system, there may be additional steps to sync your classes with Code.org:

• Click here for steps to setup your classes with Google Classroom
• Click Here for steps to setup your classes with Clever

Become a Verified Teacher: Lesson plans and levels have additional resources and answer keys for Verified Teachers, which is quick process that verifies your position at an educational institution. Click here to complete a form and you should have access to verified teacher resources in ~1 business day.

Professional Development: We recommend completing the free Exploring Generative AI Self-Paced Professional Learning Module before starting the unit. The module is designed to take 2-3 hours to complete. No previous computer science or AI experience is required.

Before beginning the unit, please have students complete a short 5-question presurvey. When students finish, they can start the warm-up in the next level.

Although their answers are anonymous, the aggregated data will be available to you once at least five students have completed the survey.

Before starting the unit or while students take their survey, please complete this teacher pre-survey. Your feedback is valued and appreciated!

LEADERS Framework: Throughout this unit, there are opportunities for students to discuss the impacts of generative AI on society, both the intended and unintended consequences. Navigating these conversations so they elevate and honor the experiences of your students, while also keeping the conversation grounded and connected to the lesson content, requires careful consideration and intentionality as the discussions are happening.

To help prepare for these moments, we recommend reviewing the LEADERS Framework for facilitating lessons and discussions involving the societal effects of AI, including bias and ethical considerations. Not all of the tenets are applicable to every lesson, but many of the teaching tips and discussion goals in this lesson follow the principles outlined in this framework.

Getting Familiar: These are the levels students will complete throughout the lesson. We recommend briefly exploring the levels yourself before reading the rest of the lesson plan. Some levels are meant to be completed independently by students, but others are meant to motivate class discussion or lead to moments of direct instruction in the slides. Very rarely should students be working through levels completely self-paced.

Teacher Exemplars: Verified teachers can access exemplar responses and "what success looks like" tips for each level in the Links section of the lesson plan. If the resource doesn't appear for you, you may need to become a verified teacher - learn more about this process here.

Teaching with Student Responses: Certain level types are designed to be leveraged with our Student Summary feature, specifically: Predict & Discuss and Mid-Lesson Check-In levels. You can read more about how to leverage student voices during these levels in the Teaching Guide - Incorporating Student Responses in Lessons

Teaching Tip: More Models! - To really emphasize this point, you could consider showing students the Transformer Model Page on Hugging Face. Hugging Face is a company that supports hosting and documenting machine learning models - this link should take you to a pre-filtered list of Transformer models, which are the models used to power chatbots. Exploring the list should demonstrate there are more models than students will likely have seen before, and some of the models are ones students may have heard of in the news (for example, Meta hosts all of their open-source Llama models on Hugging Face). The key takeaways here are: there are many different models not just the "most famous ones", and many of them are made available for public use and experimentation instead of being hidden or gated by major companies.

Teaching Tip: Building Chatbot Fluency - This is a great (and early!) opportunity to make sure students become fluent with using the chatbot interface and can identify important aspects of the level. Make sure students know where the instructions panel is, can send messages, know how to clear the chat, and know how to copy the chat. Building this fluency early will make later lessons easier.

Teaching Tip: Not Every Flagged Message is Inappropriate - When a student's on-task message is flagged, it doesn't necessarily indicate inappropriate behavior; it may simply be due to a specific word triggering our filters. We encourage trying alternative words or questions to re-phrase the prompt. Context is crucial, as legitimate discussions can sometimes be flagged. As a teacher, you can view student chats to make your own judgements on whether or not a chat deserved to be flagged.

Teaching Tip: Connecting Back to a Previous Lesson - If you taught the Foundations of Generative AI unit, the lesson on the Output layer involved generating a mad-lib and noticing that the computer-generated mad-lib was a lot less interesting and creative than student-generated mad-libs. This lesson gives us the language we need to describe that situation:

• The computer-generated mad-lib in that lesson uses a model with a low temperature, which means the responses are more deterministic and less random.
• If we wanted the model to be more creative, we would want to set it to have a high temperature, which means it would pick some of the less-likely choices from the class and result in a more creative (and sillier) mad-lib. If students need a refresher on the output layer of a language model, consider showing the Generative AI: Output & Probabilities video.

Assessment Opportunity: Formative Assessment - Students should connect that higher temperatures can be better in more creative, informal situations while lower temperatures can be better in more structured, high-accuracy situations. After students answer this question and as they work on the remaining levels, try to check how students answered this question so you can decide if you need to make this point directly with students. Check the Generative AI - Teaching Guide for Student Responses for more information about how to use Code.org’s teacher dashboard to view this data and potentially differentiate your instruction with students.

Teaching Tip: Higher Temperatures can get Flagged as Inappropriate - Because temperature increases the "randomness" of responses, some responses back from the chatbot may get flagged by our content moderation system as inappropriate. In particular, asking the chatbot to summarize "Of Mice and Men" with a temperature of at least 0.8 will often be flagged as inappropriate. If this happens, consider making it a teachable moment that addresses two important topics: (1) that higher temperatures lead to more random responses, and this response happens to be so random that it's being flagged by our content moderation system, and (2) messages sent back from the chatbot can also be flagged as inappropriate, so students shouldn't try to get the chatbot to respond with inappropriate comments and if a legitimate conversation gets flagged as inappropriate, they may need to lower their temperature.

Teaching Tip: Tracking Hallucinations - Consider establishing a structure for students to track hallucinations they find throughout the lesson, similar to a “mistake chart” or “bug chart” in math or computer science classes. In general, if you have any normalizing and celebrating error routines, they can be playfully expanded here as well to help students pay attention to hallucinations throughout the unit.

Ethics Opportunity: Real-World Connections - This article connects generative AI with news reporting and determining factual information. Consider having students do some additional research on recent events that may have involved factually incorrect information influenced by generative AI and the consequences of that information making its way into the mainstream consciousness. Consider this in the context of "misinformation" or "viral trends" that are actually AI generated. As students consider the role of generative AI in our information landscape, also prompt them to come up with solutions or ways to address these problems - this aligns with the "Show Solutions" aspect of the LEADERS Framework

Teaching Tip: Key Vocabulary - The vocabulary from each lesson is provided at the end of the slide deck. This can be useful to print and keep available for students throughout the unit, or this slide can help facilitate your own vocabulary strategies. One possible strategy is to build a word wall throughout the unit, providing both a written definition and a visual that students can reference as-needed during the rest of the unit.