Campfire Session — Using Chats 101

Introduction • 00:00

• Lulu Gao introduces the session and mentions how last week's session on Flint activities goes hand-in-hand with this one.

• She also covers the session agenda and expectations.

Ice-breaking news • 00:55

• Lulu brings attention to the newest Anthropic Education report on How Educators use Claude.

• She highlights the report's findings on how teachers are using a chat tool like Claude to automate tedious tasks, collaborate as a thought partner, and personalize learning experiences for students.

• She also mentions how teachers are creating their own AI tools for themselves and for students—much like how Flint activities work, but requiring more AI expertise from the educators.

About chats • 04:06

• Lulu distinguishes chats from activities, clarifying that chats are one-off AI conversations for quick ideas, while activities involve designing the AI's behavior for student interactions. A recurring conversation approach is described, using a consistent assistant for tasks like writing session descriptions and background material, to streamline workflow.

• The session covers the rationale for using Flint chats, highlighting privacy protections, language capabilities, coding, whiteboarding, and image generation as built-in tools, and notes ease of sharing chats and activities within the school.

• A comparison between Flint and ChatGPT is shared.

• The demonstration shows Flint's whiteboard and chat differentiation, enabling students to draw graphs and annotate images, while teachers access planning and feedback tools. The assistant Sparky provides stepwise guidance, and privacy controls allow administrator oversight of chats.

• Student vs teacher chats are discussed, including how Flint treats the two roles with different levels of helpfulness and how topic recommendations within the platform for both teachers and students differ.

• Tim asks about student registration and vetting, prompting explanations of teacher versus student roles and initial default student status. Lulu Gao clarifies administrator-led role designation and post-join privileges.

• Two notable topics are discussed: AI detection and visibility into student work within Flint. The discourse emphasizes reliability concerns with AI detection, which has been shown through research to have potential biases against non-native English speakers. Flint encourages observing student interactions over policing AI usage.

• A workspace is created for a user, with guidance on signing in and accessing tools. It is explained that a free start is available and administrators have control over the workspace, including chats and analytics.

• Features and limitations to note with Flint chats are summarized and some are demonstrated.

• An example of graphing vs image generation in Flint is shown, so educators can better troubleshoot when visuals in Flint contain inaccuracies.

Chat examples • 21:56

• A detailed walkthrough demonstrates creating a worksheet and how Flint can generate and modify math content, including equations, with options to export and reuse across platforms. The presenter showcases quick generation of practice questions and the ability to tailor outputs, including two versions for different student needs.

• A chat example of creating a rubric is demonstrated. Teachers can ask Flint to create differentiated rubrics based on students' knowledge level, English proficiency, and even include translated rubric criteria to make the expecations more accessible to students and their families.

• A feature is demonstrated where a rubric can be attached to an activity in Flint, enabling direct use in student evaluation with rubric guidelines. The process includes selecting the rubric, attaching from Flint’s existing attachments, and ensuring alignment with rubric points.

• A chat example of using Flint for email writing is demonstrated. Lulu shows how you can speak directly to Flint to speed up the process of generation.

• A fourth example of a chat generating questions based on content upload is shown. Lulu also shows how teachers can take those questions and directly make an interactive Flint activity with them.

• A final chat example of using Flint as a project planning partner is demonstrated. Flint can generate the lesson plans. rubrics, and more and educators can brainstorm with Flint how to incorporate more Flint activities into the project that support students, push their creativity, and show their thinking process better.

• Memories and personalization are explained, including the ability to remember teacher preferences and to review or delete stored memories. A short demonstration of updating the user profile and memory management is provided.

Conclusion and final Q&A • 49:30

• The group discusses using Flint with guardrails and monitoring. Administrators flag inappropriate messages and teachers can see student responses, which deters misuse.

• A plan to organize future sessions emerges, including posting links, calendars, and sharing QR codes. Timely follow-up and community engagement are emphasized.