Artificial Intelligence (AI)

AI Turned My Academic Journal Article Into An Engaging Podcast For Social Media Pros In Minutes with Google’s NotebookLM.

~ K. Quesenberry ~ Leave a comment

 I recently published academic research in the Quarterly Review of Business Disciplines with Michael Coolsen titled, “Engagement on Twitter: Connecting Consumer Social Media Gratifications and Forms of Interactivity to Brand Goals as Model for Social Media Engagement.” Exciting right?

If you’re a research geek or academic maybe. A social media manager? No way. Yet, I know the findings, specifically our Brand Consumer Goal Model for Social Media Engagement is very exciting for social media pros! So I wanted to write this blog post.

But, as you can tell by the title, an academic audience, and a professional audience are very different. Taking a complicated 25-page academic research article and translating it into a practical and concise professional blog post could take me hours.

I’ve been meaning to experiment with Google’s new AI generator tool NotebookLM so I thought I would try it. Thus, this blog post is about our research on a social media engagement framework and how I used AI to streamline my process to create it. As a bonus, I got a podcast out of it!

My co-author and I did the hard work of the research. I was okay with an AI assistant helping translate it into different media for different audiences. Click for an AI Task Framework.

Using NotebookLM.

Our study was on types of content that generate engagement on Twitter, but the real value was a proposed model for engagement. So before uploading any of the research into the AI tool, I condensed it to just the theoretical and managerial implications sections. Then I added a title, the journal citation, and saved it as a PDF.

NotebookLM uses Gemini 1.5 Pro. Google describes it as a virtual research assistant. Think of it as an AI tool to help you explore and take notes about a source or sources that you upload. Each project you work on is saved in a Notebook that you title. I titled mine “Brand Consumer Goal Model for Social Media Engagement.”

Whatever you upload NotebookLM becomes an expert on that information. It uses your sources to answer your questions or complete your requests. It responds with citations, showing you original quotes from your sources. Google says that your data is not used to train NotebookLM, so sensitive information stays private (I would still double-check before uploading).

Source files accepted include Google Docs, Google Slides, PDF, Text files, Web URLs, Copy-pasted text, YouTube URLs of public videos, and Audio files. Each source can contain up to 500,000 words, or up to 200MB for uploaded files. Each notebook can contain up to 50 sources. If you add that up NotebookLM’s context window is huge compared to other models. ChatGPT 4o’s context window is roughly 96,000 words.

When you upload a source to NotebookLM, it instantly creates an overview that summarizes all sources, pulls out key topics, and suggests questions to ask. It also has a set of standard documents you can create such as an FAQ, Study Guide, Table of Contents, Timeline, or Briefing Doc.

You can also ask it to create something else. I asked it to write a blog post about the findings of our research. You will see that below. Yet, the most impressive feature is the Audio Overview. This generates an audio file of two podcast hosts explaining your source or sources in the Notebook.

The NotebookLM dashboard gives you a variety of options to interact with your sources.

Using Audio Overviews.

There are no options for the Audio Overview so you get what it creates. But what it creates is amazing! My jaw literally dropped when I heard it. And it will give you slightly different results each time you run it.

I noticed things missing in the first audio overview such as the journal and article title and the authors’ names. I did figure out how to make adjustments by modifying my source document. Through five rounds of modifying my source document, I was able to get that information in and more.

Sometimes overviews aren’t 100% accurate. It says, “NotebookLM may still sometimes give inaccurate responses, so you may want to confirm any facts independently.” In our research article we give a hypothetical example of a running shoe brand following our model. It was not real. But in one version of Audio Overviews, the podcast hosts talk as if the company did what we said and got real results that we measured.

I was impressed that in other versions it didn’t use our example and applied the model to new ones. One time it used an organic tea company and another time a sustainable clothing brand. On the fifth attempt it even built in a commercial break for the “podcast.” This last version gave my running shoe example and added its own about a sustainable activewear brand.

What’s really interesting about the last version is that it pulled in other general knowledge about social media strategy and applied it to the new information of our study. At the end, the hosts bring up how our engagement model will help know what to say but that social media managers still need to customize the content to be appropriate for each social platform. That’s a social media best practice but not something we mention in the article.

The Audio Overview Podcast NotebookLM Created.

 

It’s amazing these podcast hosts discussed our research and explained it so well for social pros. What’s more amazing is that they are not real people! Yet NotebookLM did more. Below is the blog post it wrote. It included our diagram of the model, but had trouble getting it right. So, I replaced the image with one I created from our article.

Brand Consumer Goal Model for Social Media Engagement.

This post examines a model for social media engagement based on an October 2024 study in the Quarterly Review of Business Disciplines. “Engagement on Twitter: Connecting Consumer Social Media Gratifications and Forms of Interactivity to Brand Goals as Model for Social Media Engagement,” published by Keith Quesenberry and Mike Coolsen.

The Brand Consumer Goal Model for Social Media Engagement is a framework to help social pros create more effective plans by aligning brand goals with consumer goals. It emphasizes understanding the motivations behind consumer engagement and tailoring content accordingly.

How the Model Works

The model outlines three key brand goals:

  • Building brand community (Reach): This goal focuses on expanding the brand’s audience and increasing awareness.
  • Building brand-consumer relationships (Response): This goal aims to foster brand interaction and engagement.
  • Building brand-consumer bonds (Recognition): This goal seeks to create brand emotional connections and loyalty.

Each brand goal is associated with a corresponding consumer goal and form of social media engagement as seen in the graphic below. Consumers can manage people’s impressions and persuade others by sharing certain brand posts. They can gain information about a brand with other types of brand content for deliberation. Then they can bond with others regulating their emotions in evaluating brand posts with other fans.

Brand Consumer Goal Model for Social Media Engagement Template
Click on the graphic to download this model for social media engagement.

Here is an example

  • A sneaker brand launches a running shoe and aims to build brand community (reach) creating content that appeals to runners’ desire for impression management and persuasion. This could be sharing inspirational stories about runners breaking records encouraging retweets and brand visibility.
  • Once awareness is established, the brand could shift its focus to building relationships (response) by providing information about the shoe’s features and benefits, appealing to consumers’ information acquisition goals, and prompting replies and comments seeking further details.
  • Finally, the brand can foster brand bonds (recognition) by sharing content that resonates with runners’ social bonding and emotion regulation needs, such as posts about the challenges and rewards of training, which encourage likes and emotional connections.

Key Insights from the Study

The study found that simply using popular content types like videos or photos is not enough to guarantee success on social media. The message delivered with the content is crucial.

Marketers need to consider:

  • The target audience’s motivations for using social media
  • Buying cycle stage (awareness, consideration, purchase, loyalty)
  • Desired word-of-mouth function (sharing, deliberation, evaluation)

The Brand Consumer Goal Model for Social Media Engagement offers a strategic framework for developing effective social media campaigns. By understanding the motivations behind consumer behavior and aligning content with both brand and consumer goals, marketers can achieve better results and build stronger relationships with their target audience.

I hope you found this look at NotebookLM and the insights from our social media research helpful. In what ways do you think NotebookLM can help in your job? In what ways can the insights from the Brand Consumer Engagement Model improve your social media content strategy?

NotebookLM Could be a Great Study Tool for Students.

NotebookLM could be a great tool for student learning if used as a study guide, reinforcement, or tutor. It would have a negative impact if used to simply replace reading and listening in the first place. What’s missed when you use AI in the wrong way is depicted in the graphic below. It is from a previous post on the importance of subject matter expertise when using AI

Personally, I was fine using this tool in this way. My co-author and I did the hard work of the research. This AI assistant simply helped us translate it into different media for different audiences.

This graphic shows that in stages of learning you go through attention, encoding, storage, and retrieval. You need your brain to learn this process not just use AI for the process.
Click the image for a downloadable PDF of this graphic.

Half of This Content Was Human Created!

UPDATE: Google has released a new version of NotebookLX where you can customize the Audio Overview before processing. I was very impressed with this feature. For example, I had another academic article published about a new no tech policy in the classroom that I implemented after COVID restrictions were released. I uploaded the academic article and before processing I Customized the Audio Overview telling NotebookXL that my target audience was college students distracted by technology in the classroom and to keep the overview shorter for their short attention spans. Here is the result:

Joy Interrupted: AI Can Distract From Opportunities For Learning And Human Connection.

An image of a poster promoting the Ross Gay even on a college campus.
~ K. Quesenberry ~ Leave a comment

This is the fifth post in a series of five on AI. My last post discussed why we need more than prompt engineers, but also subject matter experts. This post discusses the danger of losing that learning if AI is not used in the best ways.

Last spring, I went to a campus reading from New York Times best-selling author and poet Ross Gay. It was well attended by faculty, administrators, employees, community members, and students. In front of me, three students sat with two on laptops and one on a phone. My first thought was a professor required attendance and they were taking notes for an assignment.

A great opportunity on campus was this author’s reading and book signing.

 

Ross began drawing me in with engaging stories of happiness and sorrow and simple delights found in life if we pay attention. He’s a master observer of joy found in everyday moments. His message and delivery were powerful, yet, my attention was soon distracted by the busy screens in front of me.

Glancing down it was obvious the students weren’t taking notes. They didn’t look up at the author at all. The student in the middle was watching a video on the phone. The two on laptops were jumping back and forth between different websites, documents, emails, and social media.

Despite Ross’s dramatic reading, I had trouble focusing with three screens flitting around in front of me. I imagined what it’s like to be a student in the back of a lecture hall or even a small classroom with dozens of student multitasking screens in front of them.

AI promises to free us from busy work.

Between the author’s readings, I glanced down again, hoping to find evidence of something related to the event and this author. Instead, I noticed the ChatGPT screen. Maybe the student was using it to supplement an assignment or get help with a difficult task. Perhaps I’d see how a professor integrated AI into a class.

Instead, I saw quiz questions from the university learning management system. Each question and answer was quickly copied back and forth between ChatGPT and a course quiz. Twenty questions were answered in less than a minute. I saw no effort to answer questions first or even read them. Did the student not know this was wrong? Or were they so engrossed in the screen that they forgot their surroundings? Perhaps the student views quizzes as busy work, not a learning tool to ensure reading and internalizing information.

AI promises to free us from drudgery to explore human creativity and imagination. The article “How AI can save you time: 5 skills you no longer need to learn,” tells us we can now skip learning skills like writing because AI will do writing like reports and news articles for us. I wonder what creativity the journalist will explore when Euro News outsources articles to AI. I don’t want to be freed from writing. My creativity and imagination are explored through writing and evidence tells us writing is how we learn.

If the student in front of me was using AI to save time to explore human creativity, they missed one of the best opportunities that semester. While they focused on their screen using AI, a poet expressed the joy of being human moving some in the room to tears.

Sometimes there is no shortcut to learning.

Much of AI is being marketed to us and students as a shortcut. The easy way to complete a task, assignment, paper, or degree. In AI’s Promise To Pay Attention For You., Marc Watkins of Mississippi AI Institute, says, “Many third-party app developers are building off of OpenAI’s API to create apps that promise an end-to-end user experience where a machine listens for you, complies with information, and then creates bespoke summaries all so you don’t feel burdened by listening or thinking about the content.”

TikTok is full of student videos promoting these apps as the easy way to an easy grade. I’m all for removing friction to make banking, car buying, and hotel booking easy. But is easy the best way to learn? What if friction and struggle are how we learn? In an op-ed Jane Rosenzweig of Harvard College Writing Center says, “Our students are not products to be moved down a frictionless assembly line, and the hard work of reading, writing, and thinking is not a problem to be solved.”

If not used properly AI can get in the way of learning. This summer I received an email marketing assignment in which a student “wrote” bland generic email copy. Then a paragraph explained how the email “fosters a deep emotional connection with the audience” and “reflects a deep understanding of the target audience’s needs.” But it didn’t! It sounded like the correct but unfeeling, general copy LLMs tend to generate.

The LLM knew what good email copy should do, but couldn’t write it. My student needed to be the human in the loop. I can teach how to write copy that forms an emotional connection with the audience based on human insight, but not if a student uses AI to write the entire assignment. Why would an employer hire them if AI could complete the entire project on its own?

Is liberal arts education the answer to AI job losses?

If AI takes away skills, many say the way to remain relevant is through liberal arts. Business Insider says AI startup founders hire liberal arts grads to get an edge. In Bloomberg, a Goldman Tech Guru says AI is spurring a “Revenge of the Liberal Arts.” WIRED proclaims, “To Own the Future, Read Shakespeare.”

IBM’s AI chief advises students who want a tech job to learn the language and creative thinking skills you get in the liberal arts. Because AI speaks our language not computer code we need prompt engineers “to train it up in human behavior and thinking.”

Marketing AI Institute’s Paul Roetzer believes the next generation entering the workforce will remain relevant with a broad-based liberal arts education. Literature, philosophy, history, and art are what make us human and teach critical thinking, analysis, creativity, communication, collaboration, integrity, understanding, and nuance. What AI can’t do.

But what if AI can also get in the way of learning liberal arts? Using AI to skip the reading and skip the writing skips the learning to save time for what? Doing the reading, processing the information, committing it to memory, and explaining it through writing is how you learn critical thinking and creativity. To have an imagination you need knowledge.

Is AI the answer to our loneliness epidemic?

In 2023, the Surgeon General released an advisory warning of a crisis of loneliness, isolation, and lack of connection. Nearly half in the U.S. experience loneliness which can increase the risk of death comparable to smoking. Rates of anxiety and depression on college campuses have never been higher. More than 60% of college students report at least one mental health problem.

Some are promising AI can solve this problem with Artificial intelligence friends. I sat in a room full of people with an author talking about human connection while students in front of me focused on their screens. Are AI friends really the best solution? What I often see in the classroom is students not talking to each other because they are focused on their screens.

After Ross finished there was a Q&A. A mental health professional behind me wanted to thank him. She gives her patients struggling with anxiety and depression Ross’s books as homework. Many report back that the books make the difference in being able to get out of bed some days.

I did not miss the irony. Many students struggle with anxiety and depression and many feel screen time is a part of it. Yet, here I was between a mental health professional, students, and one of her solutions. A real human in the room. I’m sad for the students who missed out on the joy of the evening – right past the screens taking their attention. In reading Jonathan Haidt’s The Anxious Generation my understanding and empathy for Gen Z has grown.

The exception rather than the rule.

It’s important to note this was a couple students. I’m grateful for the larger group of enthusiastically engaged students at the event. In my experience, most students are not looking for the easy way out and want to learn their disciplines by integrating AI in beneficial ways. But they need our guidance.

In a review by Turnitin, they found that of the more than 200 million writing assignments reviewed by Turnitin’s AI detection tool last year, some AI use was detected in just 1 out of 10 assignments. Only 3 out of every 100 were generated mostly by AI.

Education experts warn that focusing too much on AI cheating can cause distrust between instructors and students. We should frame the conversation around ways AI can both support and detract from learning. Our role is AI literacy providing specific guidance on when and when not to use AI.

I hope to educate students on the role of AI in their lives and how to make intentional choices about what to outsource to AI, what to keep for ourselves, and how to prepare for careers with AI to keep humans in the loop.

This Was Human Created Content!