🤖 The good and the bad about Gems and NotebookLM in Google Classroom

🎙️ In-person workshop: Student Writing in the AI Age

I’m trying something new: an in-person workshop down the street from my school!

Title: Student Writing in the AI Age
Description: When ChatGPT can do the writing, how do we keep students thinking? Get plenty of real talk and practical strategies here.

Date: Monday, November 17, 2025
Time: 8:30am to 3:30pm
Place: The Inkwell, 114 N Jefferson St, Rockville, IN 47872
Cost: $99 (includes lunch on-site) (group discount pricing available)

If you’re from Indiana or Illinois (or the Midwest): You might be close enough to consider this! Would love to share the day with you working through this topic.

If you’re NOT from Indiana or Illinois: I’m planning on turning this workshop into an asynchronous online course in 2026. I’ll keep you posted!

Note: Please only register if you’re planning on attending in-person.

PS: Are you interested in bringing me to your school, district, or even to run this workshop? Hit reply and let’s discuss!

📚 New AI resources this week

1️⃣ Rising Use of AI in Schools Comes With Big Downsides for Students (via EdWeek) — A new report raises serious concerns about the potentially negative effects of AI use on students.

2️⃣ School Districts Are Testing AI, but Few Have a Clear Strategy (via Government Technology) — A new survey of school IT leaders highlights strong interest but ongoing risks.

3️⃣ 5 keys to AI success: A roadmap for K–12 administrators (via eSchool News) — AI’s growing presence can make teachers nervous, but it doesn’t change the fundamental mission of education.

📢 Your voice: AI updates in Google Classroom

Last week’s poll: How do you feel about the AI updates in Google Classroom?

🟨🟨🟨⬜️⬜️⬜️ Very optimistic (17)
🟩🟩🟩🟩🟩🟩 More optimistic than pessimistic (37)
🟨🟨🟨⬜️⬜️⬜️ Neutral (equally optimistic/pessimistic) (21)
⬜️⬜️⬜️⬜️⬜️⬜️ More pessimistic than optimistic (2)
⬜️⬜️⬜️⬜️⬜️⬜️ Very pessimistic (1)

More optimistic than pessimistic: While my initial excitement is for sure more optimistic, and I think there is hope that these tools will be amazing tools for educators, the technical specs are concerning. […] I'm maintaining optimism, because Google usually gets there eventually, but also am hesitant to give it a "yes" for our educators until we know more... — A. McCrumb

Very optimistic: This is a great way for educators to introduce AI into the classroom and integrate it into assignments when they deem it appropriate for students. Additionally, the studio creations in NotebookLM are great ways for educators to take existing curriculum materials and create additional resources to help make learning more multi-modal. — S. Fahey

Very pessimistic: I am a HUGE supporter of the use of AI in the classroom, but I think we need more control over what students are getting access to without prior learning. Many only see it as a better Google Search, and that's not what it is at all/should be used for (for the most part). — M. Hay

Neutral (equally optimistic/pessimistic): Being techy, I love the idea, but my teachers are still learning. We have been flooded with AI, and it's difficult to keep up with all the changes and then spread that info to the teachers. — T. Applegate

Neutral (equally optimistic/pessimistic): Concerned about students using these tools without the proper supports and training in the classroom. This leads to students using them on the sly and perhaps to lessen their cognitive load. We want them to use their brains and then the tools in a coordinated and smart way, but when they simply appear without instructional design addressing the use, it will likely lead to bad habits. — S. Judge

🗳 Poll: Teacher oversight in Google Gemini

Instructions:

1. Please vote on this week’s poll. It just takes a click!

2. Optional: Explain your vote / provide context / add details in a comment afterward.

3. Optional: Include your name in your comment so I can credit you if I use your response. (I’ll try to pull names from email addresses. If you don’t want me to do that, please say so.)

💎 Gems and NotebookLM in Google Classroom: The good and the bad

On the surface, the news sounds fantastic …

Google Classroom is getting Gems — Google’s customizable AI chatbots — and NotebookLM — its powerful research assistants that create those viral custom AI podcasts and more.

Last week, we covered the basics — what Gems and NotebookLM are all about and why it’s a big deal. (Read about that here if you missed it.)

Today, it’s time for critique and judgments … the good and the bad!

Mostly, I’m optimistic about this. Honestly, having powerful custom-created chatbots natively in a learning management system is a combo I’ve been waiting for.

But there’s also potential for problems!

Let’s dig into both the good and the bad …

The good: Gems are a potential gamechanger.

Gems in Google Gemini (gemini.google.com) are AI chatbots you can give your own custom instructions. And now, you can share your custom Gems (chatbots) directly to students in Google Classroom.

Imagine the implications: The teacher designs a chatbot and assigns it in Classroom. And the chatbot isn’t a third-party app acting as the middle man for an AI model. It’s Google Classroom and Google Gems … all under one roof.

If teachers can learn how to write good instructions for Gems — and share them in Classroom — there’s a ton of potential here. Instead of turning students loose in a large language model (LLM) like Gemini or ChatGPT on their own, the teacher can set the guardrails and design the task. I love that.

The good: NotebookLM is such a powerful learning tool.

NotebookLM (notebooklm.google) was the lifeline that I tossed my own two daughters (college sophomore and high school senior) when they were struggling with some classwork.

If you’re not familiar, I wrote about NotebookLM and its uses in the classroom here.

My college sophomore has a professor whose lectures don’t make much sense to her … but when she throws the lecture video transcript AND the lecture slides into NotebookLM, she can get answers and explanations that do make sense to her.

The ability of NotebookLM to take any sources you give it — PDFs, websites, audio, files in Google Drive — and answer questions is huge.

But throw in NotebookLM Studio, where it’ll make audio overviews (aka “podcasts”), video overviews, mind maps, and other custom documents? It becomes the personal tutor and learning support so many students have craved — and desperately needed.

The good: Ease of use

It’s right there! Just click the Classwork tab and click “Create.”

When it’s enabled and available, you’ll be able to create and share a Gem — an AI chatbot with instructions that are just right for your student(s).

Even better: You can use a template to help you get started. That way, you don’t have to come up with all of the instructions yourself.

No outside apps.

No “go here and create and share a link with your students.”

Just a few clicks in the learning management system you already use.

The good: Potential mass personalization of learning at scale

The implications of this are fantastic.

In my Spanish classes, I could (theoretically) create custom Spanish tutors that interact with my students where …

• they use the vocabulary words that I choose

• they use the specific grammar structures we’ve actually covered

• they describe concepts using phrasing and explanations that I commonly use

• they pull in things that interest my actual students — and places in our local community to make it feel familiar

(Note: With pretty much all AI assistants, I’m still struggling to get a really good world languages chatbot that interacts with students in Spanish effectively. Still tinkering, but actual results as of today haven’t been great.)

This goes for just about any grade level and content area. If your students are old enough to read and type, you’ll be able to create custom chatbots to interact with them. (Whether you should is a different issue. More on that in a moment.)

This is huge. And generally, I’m really, really excited.

But with anything AI in edu, we have to proceed with caution. For example …

The bad: Google is a behemoth, and we’re feeding it

This whole idea of free AI tools for schools inside of Google sounds like a no-brainer win for everyone.

But I have to keep reminding myself of something related to Google’s business model …

They want to hook you in and keep you as a customer as early as possible. And that effort starts, in many schools and districts, when kids are kindergarteners. Five or six years old.

Just remember that Google is a behemoth (Microsoft, too) and that they can afford to offer free tech to schools … but they’re not doing it for philanthropy. There’s an end game.

“If I can do everything with Google, why would I ever use anything else?”

It feels right and good and honorable … until you think about it for a while.

Not saying this is a reason not to use Gems and NotebookLM and Google Classroom.

Just … eyes wide open, OK?

The bad: There’s a disconnect between Gemini/Classroom and teachers

When Google announced some of its big AI moves with Gemini in education this summer, I wrote about some of the concerns I had about it.

Specifically: I wish they would have connected teachers to their students’ AI use more closely.

• The only way teachers can access the transcripts of student interactions with Gemini is through the admin console … and teachers don’t have direct (or easy) access to that.

• Teachers don’t get any insights or AI analysis of student AI interactions … which is a big missed opportunity to put valuable data in teachers’ hands that could inform instruction.

• If students have inappropriate conversations with Gemini — or suggest that they’re going to harm someone or themselves — there’s no mechanism to alert teachers immediately.

• Plus, Google Classroom has AI teacher tools to create teaching resources … but again, the ability to create data-informed resources is missing because insights from student AI interactions are disconnected from the teacher side.

What I see as a shortcoming of Google’s AI in education is actually a designed feature of many K-12 AI apps like Brisk Teaching, SchoolAI and MagicSchool. You get instant, actionable suggestions to adjust instruction from student interactions. Plus, teachers find out right away if harmful or inappropriate stuff is happening with students.

I’ve seen people argue both sides of this incessantly on LinkedIn. “We don’t need to surveil students,” some have said. “Spying on student activity all the time is oppressive.”

I get that. And they’re right.

I’m not as passionate about reviewing transcripts of student chats with AI.

It’s the stuff that can inform instruction — and keep students safe — that I’m passionate about.

The bad: It’ll be available for students who aren’t ready for it yet

Google Gemini and NotebookLM are available to students of all ages. 

If it gets turned on and students have access to it, we’re opening up a direct line from children to an AI model.

In many cases, this will be just fine. For students who understand how AI works — its strengths and its limitations — it’ll be fine. If students are old enough to know that AI isn’t human and know how to adjust based on that, it’ll be fine. (I wrote about concerns about students developing unhealthy relationships with AI — and what we can do.)

It’s the students who don’t understand those distinctions. Those are the ones I’m worried about.

AI literacy is a huge topic in education circles right now. As AI becomes increasingly widespread, it’s important for all of us — children AND adults — to understand how it works and what’s at stake.

Some schools and districts have been very proactive, providing AI literacy training to students and teachers. They’re wielding this powerful-yet-dangerous tech tool as responsibly as possible.

It’s the schools that don’t — but still turn on AI tools — that I’m worried about.

Not all students are prepared to interact with AI yet.

Couple that with an AI ecosystem where teachers are disconnected from student use? I’m worried.

The bad: AI can hurt education when not used effectively

Let’s move away from the dangers that unfettered student use could pose.

Now, let’s talk about learning.

I haven’t met a teacher who hasn’t had some level of concern about the impact of AI on academic integrity.

(There are some AI influencer cheerleaders who don’t seem to have those concerns, but I don’t think they remember what it’s like to teach in a classroom anymore.)

There’s real merit to these concerns. AI can do the thinking for students. It can shortcut the cognitive struggle that students need for learning to truly happen.

If we design learning intentionally — creating activities and lessons where AI is used at the right time for the right reasons — we can maximize the benefits and minimize the drawbacks.

Unfortunately, not a lot of teachers know how to do that effectively quite yet.

Thankfully, Google has mitigated this concern somewhat with LearnLM, the AI model that powers student interactions with AI. LearnLM is trained to interact with students based on learning science principles and pedagogical best practice. According to its website, it focuses on:

• inspiring active learning

• managing cognitive load

• adapting to the learner

• stimulating curiosity

• deepening metacognition

Honestly, if it pushes for all of those things, then it’s helping to fill a void that many teachers don’t include in their very traditional teaching practices.

That’s a step in the right direction. But I still think we need to be careful when putting students in front of AI models that very ambitiously want to do things and complete tasks for us … tasks that create learning in students.

Bottom line: Tread cautiously

I know … I just spent the last half of this post talking about my concerns. I think it’s good to be cautious.

All in all, I think this is a great step in the right direction for Google. LearnLM is a firm foundation to base its AI efforts on. Gems and NotebookLM are powerful tools to put directly in the hands of students.

I think we can hold extreme optimism in one hand — and a healthy dose of caution in the other.

I’m all for making the most of these new tools — and to have them available to anyone for free? It’s a version of access and equity that we want to see. (Avoiding the “pay to play” … or “pay to learn” that causes equity gaps.)

Tread cautiously. Eyes wide open.

Make the most of these updates — and all that K-12 AI edtech has to offer — but do so with your students’ wellbeing and learning in mind.