Will Students Need Management Skills as AI Develops?

In partnership with

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 discuss why we educators may need to shift from conceptualizing AI as a tool to conceptualizing it as an employee, why this may involve a shift to teaching more management skills, and also how to use Gemini 1.5 Pro to produce high-quality quizzes and assessments from lecture audio and video.

💡 Idea of the Week:
From AI Tools to AI Employees

Over the past year, professors and university curricula have increasingly incorporated AI tools, teaching students to leverage this technology to execute tasks typically performed by humans in their respective fields. This evolution has been transformative.

However, the trajectory of AI development already suggests an imminent shift from AI as a tool to AI as an autonomous agent — in short, AI as employee.

Autonomous AI agents like Devin, developed by Cognition Labs as a "fully autonomous AI software engineer," represent this next phase. These agents are designed not merely to assist human developers but to take more ownership of complex tasks, navigating through them with minimal human intervention and with access to a wider range of software.

This shift will likely necessitate a further evolution in university education, at least in those fields where AI agents are effective like software engineering, finance, and beyond.

Mastering the use of these agents is going to become crucial.

But how do you “use” an autonomous agent? You work alongside it or you manage it.

The emerging skillset may become less and less about performing the tasks oneself and more about managing these sophisticated AI agents. Students will need to learn not only the technical workings of AI agents but also how to strategically deploy, monitor, and integrate their capabilities within broader processes.

Yet, as we stand on the brink of normalizing AI as an employee, it is also essential to consider the future beyond this phase.

While current AI agents are autonomous within their circumscribed domains, the next step will likely involve AI reaching a stage of self-management across a wide set of domains — AI not only performing tasks but also orchestrating its workflows and managing other AIs and humans.

This prospective phase raises profound questions about the role of human oversight and the ethical dimensions of AI autonomy. Reflecting on this progression — from AI as a tool, to AI as an employee, to AI as manager — we must critically assess our role in shaping these transitions.

Are these developments inevitable, or are they a path that we are choosing? How do we prepare for the ethical and practical challenges each phase presents? How do we best prepare our students?

As we train the next generation of professionals, the focus should increasingly involve developing robust leadership, strategic oversight, and ethical governance skills.

This approach will not only prepare them for the imminent future where AI agents are tools and team members but also for a potential era where AI's independence challenges our conventional views on work, responsibility, and creativity.

📬 From Our Partners: LLM Browser Integration

MaxAI.me - Smart Browsing with 1-Click AI

MaxAI.me—where AI meets your browser. Instantly enhance web browsing with article summaries, email crafting, and AI-powered web searches. Ranked #1 of the day and the week on ProductHunt.

Discover why 1M+ users trust us. Install now!

🧰 An AI Use Case for Your Toolbox:
Gemini 1.5 Pro for Quiz Creation

Today, I will show you how to create quizzes and other course content from large quantities of data, whether audio files, video files, or sets of pdfs. In particular, I will show how you can upload entire lecture recordings to generate quizzes and other assessments from them, like Google demonstrates in this sped-up gif by Jeff Dean:

But first, some background…

Background

Two months ago, I reported on Google’s release of Gemini 1.0 Ultra, a large language model (LLM) that was the centerpiece of the $19.99/month “Google One AI Premium” plan. When the model underlying Gemini was first released this past December, it was promoted as being multimodal, “which means it can generalize and seamlessly understand, operate across and combine different types of information including text, code, audio, image and video.” In my newsletter on Gemini 1.0 Ultra, I tested these capabilities to see whether they could analyze handwritten student physics assignment submissions — with mixed results.

One short week later, Google announced Gemini 1.5 Pro (these names are terrible, aren’t they?!), an experimental model that is available for free in Google AI Studio. The big difference between Gemini 1.0 Ultra and Gemini 1.5 Pro is that the latter has a massive context window. In short, it can receive, process, analyze, synthesize, etc. much larger quantitites of input data, whether text, audio, video, or whatever.

Here’s how Google expresses the giant leap they’ve taken in increasing the context window size:

Now, you might be asking: what is a token? We cover this in our new “Insights Series” (enroll via the poll below), but the short answer is that they are segments of words, and 100 tokens are roughly equivalent to 75 words. This means that Gemini 1.5 Pro can handle A LOT of text — up to roughly 750,000 words!

Just this past week, Google announced that Gemini 1.5 Pro is now available in 180+ countries via the Gemini API, so you can integrate it in more complex and bespoke workflows. (They also added a ton of new features.)

With this background in hand, let’s see now one way you can use Gemini 1.5 Pro as an educator: namely, to create quizzes from the audio or video of a lecture.

Step One: Log in to Google AI Studio

Go to https://aistudio.google.com/ on a fresh browser window (trust me), log in with your Google account, and you will be presented with the “Public Preview” of Gemini’s API, which is how developers integrate Gemini into bespoke apps and websites.

In this case, you won’t need to get into any of the details to take advantage of the power of Gemini 1.5 Pro.

You’ll first be presented with a screen that looks like this:

You can rename the experiment at the top of the page by clicking the pencil icon to the right of “Untitled prompt.”

Step Two: Upload Audio or Video, Add Prompt

The next step is to provide an audio or video file via the “Insert:” options below the name. In my case, I provided a video on applied ethics that was part of an asynchronous “Introduction to Philosophy” course I taught a few years ago during the peak of the COVID pandemic.

It may take Gemini a little while to process the file, especially if it is large. Be patient. As noted above, I recommend dedicating a new browser window to the task.

Once Gemini is done processing the file, you can paste instructions next to the file in the prompt part of the “Chat” window. Here are some basic instructions to try:

Here’s what it looks like with the instructions added (as you can see, I used ~400000 tokens for a ~23 minute video):

Step Three: Profit

Finally, press the blue Gemini button and you are off to the races. Here is the first product of the above prompt and video:

I can confirm that these questions are reasonable questions to ask about the content of the video. They reflect its content accurately and they are intro-level. I can also confirm that the answers in the key are the correct ones.

When then urged to kick it up a notch — to make the quiz more challenging and wide-ranging, incorporating the content of the video in creative ways — Gemini produced the following:

Not bad!

Give it a whirl yourself…

I will be covering more complex use cases in coming months, both in free weekly newsletters and in ✨Premium pieces. Until then, be sure to check out Google’s documentation.

🤖 Enroll in Our AI and Higher Ed
Primer Email Series

If you made it this far in the email, we have another surprise for you:

We just finished developing our new “AutomatED Insights Series”!

It consists of 7 additional emails for new subscribers that convey who we are and some crucial information that we have learned and written about in the past year. Like AutomatED, it is designed for professors and learning professionals who want actionable but thoughtful ideas for navigating the new technology and AI environment — and it will get you up to speed with AutomatED, too.

These emails will come in a sequence, with delays, on Sundays, Tuesdays, Wednesdays, or Thursdays.

Every new subscriber to AutomatED is now enrolled in it by default, but you may have signed up before we released it… Sorry about that!

Still, you can enroll yourself manually by clicking “Yes” below (note: you will not get a confirmation message). You can always unenroll starting in the second installment without unenrolling from AutomatED’s weekly newsletters.