7 Ways I’ve Used Generative AI For Alumni Magazine Projects

Over the past year, I’ve been doing a lot — and I mean A LOT — of reporting on generative AI.

In the coming months, I’ll have several bylined stories in alumni magazines that illuminate the ways that faculty are using AI in their teaching and research, how alumni are using AI in their businesses and work, and the ways that students are thinking about AI’s impact on their learning. (Oh, and that “Ten years from now” story above includes plenty of AI-based insight, too!)

All that reporting made me realize that it was time for me to think about integrating generative AI to improve the ideas, processes, and writing that I do that’s linked to print magazines — and report those results to you.

So today, I’m excited to share some of the ways I’ve used generative AI, including some of the pitfalls I’ve experienced and the surprises I’ve encountered along the way.

If you’ve used generative AI in your print magazine work, I’d love to hear about it! Hit reply and tell me all about it.

Use case #1: Transcription

Service: Otter.ai

My current grade: A-

Analysis: I’ve been championing Otter’s transcription capabilities for years now, and I still think they’re top notch. I conduct dozens of interviews every single month, and Otter turns my recordings into (almost) polished transcripts in a matter of minutes.

While it’s always been an outstanding transcription tool, Otter now offers some great summary options, which can be an excellent way to pinpoint the exact areas of a conversation you want to focus on for a story or other projects.

I’ve also shared transcripts with clients who hope to use other quotes from conversations with faculty or alumni. The summary notes can be helpful so that in-house communicators don’t have to spend too much time reading through irrelevant information.

Pitfalls: Otter’s transcriptions are not perfect! For any quotes I intend to use, I always click on the files to listen to the audio itself to confirm its accuracy. (The transcript connects every word to its exact spot in the recording, so that work is easy.)

Use case #2: Interview preparation and idea expansion

Service: ChatGPT 4.0

My current grade: B-

Analysis: After I write a story pitch or a list of interview questions, I like to ask ChatGPT some version of the question, “What am I missing?” as a way to understand my blind spots or avoid any missed opportunities.

When there are specific questions I find promising, I ask ChatGPT to dive even deeper, providing nuances or new angles. This has helped me create more robust interview question lists that I use during my conversations with sources, and it’s helped me craft more useful pitches for clients.

Pitfalls: If I ask ChatGPT to list 10 areas worth exploring, I rarely read more than one that feels like a genuinely helpful suggestion. The other ideas I’ve typically covered in my initial work, or they’re too basic or irrelevant for the topic at hand.

Even the one promising suggestion might not be that helpful on its own! But it may point me in a useful direction.

I try to be really thoughtful when I use this specific process, because recent research shows that use of generative AI can siphon away more creative, outside-the-box ideas.

Use case #3: Headline generation

Service: ChatGPT 4.0

My current grade: B-

Analysis: From my perspective, headline writing is one of the most underutilized ways of increasing the impact of your magazine. Research shows that good headlines are at least 3x more engaging than mediocre ones.

That’s one reason I recommend that editors and writers create a LOT of headlines before choosing one.

While the results are almost always worth it, it’s definitely a time-consuming process. One way to speed that process up is to ask for some help from ChatGPT, which will generate an endless list of headlines on command.

Pitfalls: To be honest, the headlines are almost always mediocre. There is a certain artlessness to them that makes every option feel…fine, I guess. There’s no spark of joy or cleverness or creativity to them. That might be fine for a headline that has to align with SEO best practices, but not for a print publication, where headlines benefit from wit and cleverness.

Weirdly, generated headlines also almost always include colons, like the AI is writing a title for an academic paper. Sometimes, after I explicitly request headline ideas with no colons, 40 percent of the suggestions STILL have colons. Is Big Colon funding generative AI efforts? It’s truly egregious.

That said, there are almost always a couple that seem to offer a promising new direction.

They’ll use a specific word that might lead to an interesting hook or a turn of phrase I hadn’t thought of. If nothing else, it’s a great way to avoid the blank page syndrome.

One other note: You can also get my one-hour Headline Bootcamp course — by far the most popular course I’ve ever offered — to support your headline generation efforts.

Use case #4: Emails

Service: Google Gemini

My current grade: D-

Analysis: I send out a lot of emails as part of my print magazine work. I send out scheduling emails (the WORST), reminder emails, source review emails, you name it.

I have a handful of templates that I use for this often-perfunctory work, but could Google’s new assistant help make it even easier?

Whew, no.

At least, not yet. While there are options to “help me write,” to expand or shorten an email, or to formalize my writing, they haven’t made this work any easier, better, or less time-consuming.

That said, I’m not giving this a failing grade because it seems like it might ultimately have potential! If the service could pull in information from previous emails, for example, I can see how it might become more valuable.

Use case #5: From newsletter to social post

Service: ChatGPT 4.0

My current grade: B-

Analysis: First, let me acknowledge here that my experience is not technically related to print magazine work. However, because it has parallels to work that you might be doing for your institution, I’ll share it!

For a while, I experimented with posting on LinkedIn. I liked sharing helpful information in short posts, but I could tell I wasn’t getting the tone quite right. So I fed some of my newsletters into ChatGPT, asked it to generate a handful of LinkedIn posts, and reviewed the results.

In some ways, they were actually pretty good! It was clear that they were closer to the more standard, upbeat LinkedIn posts that seemed to get good engagement. They used emojis, bullet points, and calls to action.

Pitfalls: The posts also made me cringe a little bit, because they didn’t sound like me. I knew there was probably a way to thread the needle, but by that point, I realized I wasn’t willing to put in the time to make these posts as good as they needed to be. I ditched this experiment.

That said, If you’ve got a print magazine and want to transform longer stories into shorter, social-media friendly posts, I suspect there is real promise here. This is an area that is worth exploring.

Use case #6: Editing and proofreading

Service: ChatGPT 4.0

My current grade: C

Analysis: It’s tough to edit your own writing.  Getting a second set of eyes (AIs?) in the form of an industrious AI bot is, in theory, a dream. I send all of my drafted stories through ChatGPT to identify weak writing, typos, and grammatical errors before I submit them to an editor. What’s not to love?

Pitfalls: As an editor, ChatGPT leaves something to be desired. Its recommendations often drain a story of personality or add unnecessary formality.

Yes, it catches some typos and grammatical errors, but certainly not all of them.

I learned quickly that while this use case might provide me with a level of confidence about the quality of a story that I didn’t have before generative AI was available, it might be unearned confidence! It’s an improvement, but it’s never letter-perfect.

Use case #7: Images

Service: DALL-E

My current grade: D- for actual value, A+ for humor

Analysis: As is true with the social media post use case, image generation is not something I’ve seriously considered for any Capstone clients.

But for this newsletter? Well, maybe!

Every so often, I ask it to generate  alumni magazine-themed art, and it is always VERY BAD.

Actually, let me amend that: it’s actually not too bad if you plan to spend less than a single second looking at it. At a glance, it’s not the absolute worst.

However, if you spend more than one second looking at it, prepare to become increasingly horrified:

.

In summary: AI, from my experiments, has a real hit-or-miss quality to it at the moment. Nothing has felt like it’s quite lived up to the initial magic of that crazy 2022 rollout — but perhaps I’ll change my tune as I wade in further.