Google-Using Institution? Stay Tuned...

Welcome to AutomatED: the newsletter on how to teach better with tech.

Each week, I share what I have learned — and am learning — about AI and tech in the university classroom. What works, what doesn't, and why.

In this week’s piece, I encourage professors to use LLMs to analyze the semester gone by, I share a free Tutorial with subscribers who refer their friends or colleagues, and I discuss the seemingly countless important AI-related updates from Google.

If you or your institution uses Google Workspace, you’ll want to read to the end.

💡 Idea of the Week:
Use Tech to Reflect

As the spring semester wraps up, it is an ideal time for us professors to reflect on our teaching methods, course design, student outcomes, and course evaluation feedback.

What went well, what did not go so well, and why?

One way to conduct this reflective process is by leveraging the power of large language models (LLMs) like ChatGPT, Claude, or Gemini. On their own and with the right prompting, these tools can be very effective at conducting qualitative analyses. In essence, they can act as thought partners. Likewise, they can be paired with other tools to conduct quantitative analyses, quickly offering insights that might otherwise go unnoticed.

For qualitative evaluation, LLMs can assist in brainstorming ways to improve courses based on student feedback and/or your own observations. Try inputting (summaries of) student comments, notes from mentor/peer reviews, or your own reflections into an LLM, along with information about your educational context. The model can then generate suggestions for new teaching strategies, pinpoint themes or general issues, and propose solutions.

Quantitatively, LLMs combined with tools like ChatGPT's Advanced Data Analysis can use statistical analyses to scrutinize anonymized gradebooks, mid-semester feedback surveys, and course evaluations for patterns and trends.

Below, I explain how to access a mini Tutorial where I show some highly effective prompts you can use to conduct a comprehensive set of quantitative analyses in 15 minutes or less…

📈 Help AutomatED Grow, Get Free Tutorial

With the summer “break” upon us, I am excited to announce a new AutomatED referral program.

We are growing rapidly and many higher educators are telling us how much they appreciate our content — we have almost 50 ✨Premium subscribers now — but we also run into professors each week who have no clue about what we offer

We are hoping you can help us spread the word, if you appreciate our content!

To make it worth your time, we have referral rewards that you can earn by referring one new subscriber to AutomatED via your unique referral link. It takes only one.

You need to refer them to subscribe by June 16th to get one of these rewards.

If you are a ✨Premium subscriber who refers someone…

…you will get a mini Tutorial that walks you through how to clean and anonymize your spring gradebook, upload it to ChatGPT’s Advanced Data Analysis along with some other information about your course, and use some highly effective prompts that I have developed to help

• find large and highly statistically significant correlations between student performance on pairs of specific assignments or on assignment averages

• determine which, if any, of your formative assignments are predictive of the summative ones (i.e. do students who get As on one tend to get As on the other)

• discover whether there any content groups or areas that students tend to struggle or succeed with

• calculate which adjustments to your assignment category weights would get your students’ grades closer to those that you envisioned when you originally determined your weights.

Not only will this help you reflect on the past semester, but it will also equip you with actionable data to enhance future courses.

If you are a non-Premium subscriber who refers someone…

…you will get free access to the entirety of our most popular recent ✨Premium Tutorial, namely my comprehensive 7000-word step-by-step explanation of how to build custom GPTs to maximize their performance and prevent misuse.

All you need to do is get your friends or colleagues to subscribe by June 16th via the below link, which is unique to you and allows us to keep track of whether you have referred anyone.

You can also input their email address for them, but then they will need to confirm that they would like to subscribe by clicking a link they receive via email.

After we have checked your referrals, you will receive the appropriate reward based on whether you are Premium or not.

📢 News of the Week:
Google I/O Brings Tons of AI Updates

Google I/O 2024 went down two weeks ago, on May 14. Like prior iterations, it is for developers who build, develop with, or deploy Google software in its many forms. It is the largest developer conference Google hosts, consisting of announcements, keynotes from the CEO and the developer team, product demonstrations, networking opportunities, and more.

Originally, I was going to report on it in our newsletter last week, but OpenAI’s announcements were more pressing. However, there is no denying that Google is closing the gap with OpenAI and by extension — given their relationship — Microsoft.

While Google announced more than 100 new features and products at I/O (click here for the brief recap videos Google released afterwards), I think these are the 10 most important ones for higher educators…

LearnLM Models, Fine-Tuned for Learning, Plus Illuminate and Learn About

Google introduced LearnLM, “a new family of models fine-tuned for learning, based on Gemini.” As outlined in their 86-page technical report, these models are grounded in their collaborations with both the AI and the EdTech communities. The models are designed to inspire active learning, manage cognitive load, adapt to learner needs, stimulate curiosity, and deepen metacognition. LearnLM will be integrated into products like Search, YouTube, and Gemini.

Beyond LearnLM, Google also announced Illuminate and Learn About (for the latter, sign up here to be an early tester). Illuminate “breaks down research papers into short audio conversations”, while Learn About “explores how information can turn into understanding by bringing together high-quality content, learning science and chat experiences.”

• Why it matters: LearnLM represents a significant step forward in using AI to support education — or, at least, a sign of a serious, unified, and comprehensive foray by Google into it. With features like adjustable AI Overviews in Search, Circle to Search on Android for solving math and physics problems on Android, and Learning coach Gems in Gemini (see below), Google aims to streamline the ability of educators to create and students to receive more personalized and engaging learning experiences. These models will also inform new generative AI experiences in Google Classroom, helping teachers with lesson planning and differentiation. Illuminate and Learn About are promising, but it is early days.

Customizable Gems in Gemini Advanced

Gemini Advanced subscribers will soon be able to create customized AI assistants, or "Gems," tailored to specific tasks and needs. This feature will enable users to describe what they want their Gem to do and how they want it to respond. In essence, Gems will be Google’s version of OpenAI’s custom GPTs. (Google already has Vertex AI Agent Builder, but it is designed for enterprise use cases.)

• Why it matters: As I have argued before, custom GPTs are a gamechanger for professors, especially now that they are free for students to use. They enable educators to personalize AI assistants for tasks such as curriculum planning, pedagogical role simulation, tutoring and student advising, research collaboration, or whatever else they need. It is a big deal that Google recognizes the power of these assistants and plans to deploy them internal to Google Workspace. This addresses many concerns that Google-using institutions have about heavily leaning into custom GPTs — now they won’t need to choose.

Gemini 1.5 Pro Now in Side Panel (or coming soon)

Gemini 1.5 Pro is now available in the side panel of Gmail, Docs, Drive, Slides, and Sheets. Since Gemini 1.5 Pro has a massive context window, this means that users can engage with a lot more of your data, tailoring its responses accordingly. If you use a personal Google account, you have access to it already, and Workspace users can get it either through Gemini for Workspace add-ons (your institution will need to opt-in) or the Google One AI Premium plan. Until then, this feature is accessible to Workspace Labs and Gemini for Workspace Alpha users now.

• Why it matters: This integration will allow professors, learning specialists, administrators, and other higher educators efficiently summarize large sets of data and emails, making it easier to manage tasks and streamline communication with students and colleagues. Educators can start using this feature immediately to handle end-of-semester emails, large document reviews, and more, enhancing their workflow and saving time. If students have access to it — depending on the institution, this is only a matter of time — it will similarly enhance their ability to analyze, synthesize, and respond to large amounts of information located in their Workspace accounts.

Progress on Project Astra

Google DeepMind’s Project Astra is developing advanced AI assistants capable of understanding and responding to complex, dynamic environments. Some capabilities will be integrated into Google products later this year.

• Why it matters: As we have discussed in the past in the context of Devin, the AI software designer agent, highly responsive, proactive, and cross-domain AI agents might revolutionize the educational landscape, providing robust support for both teaching and administrative tasks. At the very least, if they are successful in industry, then professors will need to adjust their pedagogies to reflect this fact — perhaps shifting to teaching more management skills.

Veo and VideoFX, a Video-Generating Model

Veo, a competitor to OpenAI’s text-to-video Sora, “generates high-quality 1080p resolution videos that can go beyond a minute, in a wide range of cinematic and visual styles.” Likewise, VideoFX uses Veo to “turn an idea into a video clip.” These tools, yet to be released but potentially gamechanging, follow upon the announcement of Google Vids in April. Vids aims to simplify video creation for educational and professional purposes, integrating seamlessly with Gemini and other Google tools. As Google puts it “Vids is your video, writing, production, and editing assistant, all in one. It can generate a storyboard that you can easily edit, and after choosing a style, it pieces together your first draft with suggested scenes from stock videos, images, and background music.”

• Why it matters: With Veo, VideoFX, and Vids, professors and instructional designers will be able to create immersive educational experiences and engaging video content for lectures, tutorials, and presentations with ease, enhancing the digital learning experience for students. (Storyboard mode “lets you iterate scene by scene and add music.”) Institutions should prepare to incorporate this tool into their multimedia resources for more dynamic content delivery.

Gemini Live for Natural Conversations

In the coming months, Gemini Live will introduce a conversational experience to Gemini Advanced, allowing users to interact with the AI like they would a person. This feature is Google’s response to the conversational ability of GPT-4o.

• Why it matters: Just like GPT-4o, this feature might improve real-time, interactive learning, whether in class, in tutoring settings, or at home, providing immediate assistance and feedback to students. Faculty members should reflect on ways to incorporate this sort of tool into their teaching methods to enhance student engagement and support.

Gemini 1.5 Pro with 2 Million Token Window

Developers using the API who join a waitlist and Google Cloud customers can now use Gemini 1.5 Pro with a 2 million-token context window.

• Why it matters: This capability will continue to expand the abilities of Gemini to handle complex tasks, enabling detailed analysis and synthesis of large volumes of academic content. The 1 million token context window currently available already enables many tasks impossible with other LLMs — like analyzing lengthy recordings to generate quizzes on their content — but this update takes it significantly further. For comparison, GPT-4o has a 128,000-token context window. (Google also announced context caching in the API, which improves the LLM’s ability to retain a substantial initial context despite adding a lot more subsequently.)

Gemini 1.5 Flash for Efficiency

The new Gemini 1.5 Flash model, optimized for speed and efficiency, is available now in Google AI Studio (the Gemini API) in public preview. It is suitable for high-frequency tasks like summarization and data extraction.

• Why it matters: This model helps educators quickly process large amounts of information, aiding in tasks such as grading, reviewing research papers, and preparing lecture materials. Institutions should evaluate how this model can be used to support high-volume tasks efficiently.

“Help Me Write” Supports More Languages

“Help me write” in Gmail and Docs will support Spanish and Portuguese on desktop starting in the coming weeks, with more languages to be added over time. (Previously, Google announced “Translate for me” in Google Meet, supporting over 60 languages.)

• Why it matters: This expansion aids bilingual and multilingual educators and students, facilitating smoother communication and collaboration in diverse linguistic environments. Schools should prepare to leverage this feature to enhance language inclusivity and support within their academic programs.

New Features for Gemini in Gmail Mobile App

New features in the Gmail mobile app, such as email summarization and contextual Smart Reply, will be available to Workspace Labs users this month and to all Gemini for Workspace customers and Google One AI Premium subscribers next month.

• Why it matters: These tools help educators quickly catch up on lengthy email threads and provide more accurate and detailed responses, saving time and improving communication efficiency. Professors can stay productive on the go, making it easier to manage their inboxes during busy periods like the end of the semester.

Other announcements that I have left out include the rollout of AI Overviews to Google Search to everyone in the US (“Grounding with Google Search” is already available on Vertex AI); the announcement of Imagen 3, their latest image generation model; the dozens of ways AI is getting integrated aggressively into everything Android (see #53-#76 here); the release of PaliGemma, a vision-language open model for visual Q&A and image captioning; the promotion of an online course built with MIT RAISE to equip educators with generative AI knowledge; and more.

I am hoping to interview Google’s AI Product Manager in the coming weeks to get more insights on what Google offers — and will be offering — professors and other higher educators, so stay tuned for that, too!