• Skip to main content
  • Skip to primary sidebar

Karen@KarenCommins.com

Karen Commins

Award Winning

Atlanta Audiobook Share-rator™

  • Home
  • Demos
  • Titles
  • Reviews
  • Videos
  • Blog
  • About Me
  • Contact
  • Shop

An AI Dilemma

8 May 2026

I’m excited to be speaking on Tuesday 23 June at 4pm during the Audiobook Empire Narrator Summit! I’m passionate about my topic From Arthurian Legends to WhodunIts: Creating Your Publishing Empire With Public Domain Books. I hope you’ll be able to join us, especially since we’ll include a Q&A in the session.

You may be thinking, “That’s great, but what does your presentation have to do with AI?”

This article will answer that question.

Background

When I worked for the IRS, I read Cliff Atkinson’s book Beyond Bullet Points, and it changed how I develop presentation slides,

Atkinson instructs you to create a storyboard of your discussion points. Using his process, you’re able to scale your presentation to last from 15 to 45 minutes. You can download his invaluable templates on his site

Due to this system, I don’t make ugly, boring slides consisting of black text bullet points on a white background. Instead, each slide displays a captivating image that has something to do with the idea I will address.

As you might expect, it can take some time to locate images that correlate to your talking points. I can spend weeks creating slides for a presentation.

I often have to pay to license the graphics on my slides. I use Public Domain (PD) images whenever possible, but modern concepts usually call for modern images.

AI Advances

Whether or not we want AI attributes in our software, manufacturers are shoving AI down our throats in all of our applications — often without the capability for the user to opt-out of the AI tools.

We’re increasingly bombarded with overly enthusiastic messages like:

  • “Everybody is using AI. You better get on board, or you’ll be left behind!”
  • “The AI can create what you need in seconds!”
  • “Think of how much time you’ll save!”

In fact, the TV show HACKS included just such a conversation in this week’s episode. I was glad to hear Ava’s character speak about the lack of ethics in AI tools. Most Large Language Models (LLMs) have been trained using copyrighted texts and art without permission from or compensation to the creators. Also, the tremendous power consumption required by the data centers is doing terrible things to the people in their communities and the planet.

Meanwhile, many people have embraced AI as part of their everyday life, seemingly without regard to any privacy concerns. One of our friends told me she asks ChatGPT numerous questions every day. She said it saves her time doing research. She does investigate the options the Chatbot presents her, but I suspect most people don’t.

Other people I know create every new fad meme made with generative AI just because they saw it on social media. They either don’t know or couldn’t care less about the lack of ethics and power consumption used for their little bit of fun.

Ryan Holiday discusses in his 5/7/26 newsletter how AI is destroying our critical thinking skills. By writing our thoughts, we are “focused, patient, and disciplined.” We become different people as we learn how WE think rather than blindly accepting AI suggestions or asking a chatbot to write something for us..

My Initial Forays Into AI

I will say that I have used the AI tools in Evernote and Adobe Acrobat. In both cases, my usage is akin to a semantic search of my material contained in the application.

In Evernote, I often ask the AI Assistant to find an entry among my 20K+ notes that contains a certain quote or describes a situation. I also heavily rely on its transcription tool on newspaper articles, recordings, and even my journals because it recognizes handwriting. The linked article outlines the many corrections I must make to the transcriptions and includes info about Evernote’s data privacy policy.

With Adobe Acrobat, I have interrogated its AI Assistant about the PDF of a Public Domain audiobook as part of my prep for an audiobook I was about to record. (I’d need to fully read the Privacy Policy and related FAQs before I asked the AI Assistant anything about a copyrighted PDF.) I’ve asked it for:

  • analysis of media influences in a court case
  • character descriptions, accents, and pages where they have dialogue
  • page numbers where the author referred to himself in first person
  • pronunciation list of proper names
    • In one case, I asked it how it knew that the name should be pronounced the way it suggested and not the other way that name is commonly said. The bot supplied its reasons. I did some research to find a family member and confirmed the pronunciation with them. Adobe’s AI Assistant got it right.

Here’s where my current AI dilemma comes into play.

Experiment

In my June presentation, I’ll teach narrators about using the wealth of Public Domain works, which is material on which the copyright has elapsed. Things in the PD belong to ALL of us. If the corporate overlords who built the LLMs behind generative AI had trained them with PD works, I would eagerly use every AI tool available.

Some people argue that AI advances are no different than the Industrial Revolution. Technology advances inevitably put people out of jobs.

AI is different, though, because of its origins in stolen copyrighted work. I can’t ignore that aspect.

I have to confess, though, that my strong desire to save weeks of time, possibly generating the whole series of slides with a few keystrokes, tempted me to see what AI actually could do for me.

I have a subscription to Microsoft Office 365, which excitedly assures me that its Copilot could:

Generate and edit images, posters, infographics, banners, and more: Turn ideas into pro-grade visuals in minutes. Just describe what you want to create—AI handles the heavy lifting so you can focus on the message, not the mechanics. Then, make precise, targeted edits with our AI visual editor.

It first guided me to choose a template in one of these categories:

  • Recommended
  • Pitch Deck
  • Business and Productivity
  • Research Shareout
  • Project Report
  • Marketing

Templates have been a mainstay of Office for decades. I’m a firm proponent of templates and other ways to be efficient.

However, none of the templates looked like anything I envisioned or want to use.

I started a chat in Copilot. I first asked, “How can I generate a Powerpoint presentation based on a theme without using a template?”

It responded with an in-depth series of steps to manually make a Powerpoint file, starting with a blank presentation, choosing colors and fonts, etc. In other words, I was shown the same process I’ve followed for the last 30 years.

I then asked, “I know how to manually make a presentation. I want to know how to use Copilot to generate a set of images for my slide backgrounds that all visually related to the key word I give it.”

This time, the bot gave me a much better response, saying:

Got it —you’re not asking how to design slides, you want a repeatable Copilot workflow to generate a cohesive set of background images that all “feel like the same collection,” driven by one keyword.

It offered another lengthy set of instructions and pointed to this official reference.

I wrote a very specific prompt to request an image, and the rendered graphic beautifully matched my description. Inserting it in the slide then caused Powerpoint to give me numerous design ideas about the text.

Conclusion

My experiment made me want to see whether I could develop my whole presentation this way. I haven’t decided which way to proceed: the old manual method of finding — and possibly paying for — suitable images, or the new approach of generating them. I could even choose a hybrid plan as I’ve already found some good images on stock photo site.

Hence the title of this article — An AI Dilemma.

What choice would you make if you were me? Please leave a comment!

 

SHARE VIA:
Send email Mail
Share on TwitterTweet
Share on Facebook Share
Share on LinkedIn Share
Share on Pinterest Share
Print Print

Related posts:

  1. 6 Weeks of Showing My Work
  2. Public Domain Narration Headquarters
  3. Planning Your Trip to Public Domain World
  4. Don’t wait for vacation to send a postcard

Filed Under: Business, Narrators, Public Domain Tagged With: Adobe Acrobat, AI, Evernote, Microsoft Copilot

Reader Interactions

Comments

  1. dinogami says

    9 May 2026 at 10:43 am

    I understand the dilemma. I have made GIGABYTES of PowerPoint presentations for teaching, public lectures, and more, both before and after the advent of AI. And I empathize with the time it takes to track down suitable images and content to use, and sometimes not finding them at all. I have also created from scratch my own graphics, and even animations, to use in such presentations. But personally, I respect the time, effort, and ingenuity of my fellow creatives enough that I will absolutely never, under any circumstances, use generative AI to steal from them to create “new” images or content. And I will save myself the distress at having contributed to generative AI’s environmental impacts. I am positive I’m swimming upstream, just as I long have by recycling, conserving water and electricity, etc. when the majority are doing the opposite. I personally could not care less about keeping up with the Joneses if it comes to submitting to unethical practices while they’re all off being unethical. I don’t believe that a majority opinion or practice automatically makes it the RIGHT opinion or practice. I have to ask myself why I’m feeling pressured to use an unethical tool—who is creating that pressure, and who ultimately stands to gain if I submit to it. Lastly, I’ll say I have no problem using _assistive_ AI (such as Pozotron or Izotope RX use), but I draw the line at _generative_ AI.

    Reply
    • Karen Commins says

      9 May 2026 at 1:38 pm

      Hi, Dinogami! Thank you so much for sharing your thoughtful summation!

      You’ve fortified my objections to generative AI, especially in your penultimate sentence when you asked who is creating the pressure to use an unethical tool and stands to gain from it.

      After I published the article, I told my husband that images generated in MS Copilot may look good and save me time and money. However, expediency is not a reason to betray my ethics and integrity.

      Your decision to use assistive AI tools but NEVER use generative AI is admirable and encourages me to follow your lead!

      Thanks again!

      Karen

      Reply
  2. Dani says

    9 May 2026 at 11:28 am

    Fantastic article, Karen. I don’t think there’s a right answer here, although doing due diligence is crucial, like you found with the name pronunciation. I’m using PreRead for my current project and it told me a MC had a southern accent. I ending up reading the book a second time through and found a single reference to the MC being from Chicago, not the south. So yeah, due diligence for the win. You always produce great content, Karen, and we appreciate your insights.

    Reply
    • Karen Commins says

      9 May 2026 at 1:49 pm

      Hi, Dani! Thanks for the kind words!

      Dinogami in the comment above is willing to use assistive AI but refuses to utilize generative AI due to its unethical nature. I’m drawing my line in the sand at the same place.

      As you, I, and countless others have discovered, assistive AI often is wildly incorrect. I’ve been astonished to see how much is wrong in a simple transcription from typeset material!

      Thanks again for sharing your thoughts!

      Karen

      Reply
  3. Peter Lerman says

    9 May 2026 at 1:21 pm

    I am assuming that you are or have been using the ‘royalty free’ image websites. I think they are fine. Plenty of good content.

    The AI stuff, to me, looks like AI stuff. Lots of people use it. I don’t care for it. I find much of it has a ‘plastic’ look to it.

    I am just one lone voice in the wilderness and I imagine the majority of your audience (any audience) will not really know the difference or care much one way or the other. Perhaps people are just coming to expect it.

    Peter

    Reply
    • Karen Commins says

      9 May 2026 at 1:44 pm

      Hi, Peter! Thanks so much for the note!

      I do have a subscription to Storyblocks. It’s much better for video and audio than graphics, in part because its search isn’t finely tuned.

      I also have Camtasia Assets, which uses Shutterstock. Unfortunately, Camtasia doesn’t have access to all of the Shutterstock images, and naturally the ones I want aren’t included.

      I’d have to pay for the Shutterstock images. Obviously, I’m trying to avoid that possibility!

      I agree about the plastic look of many AI images.

      Thanks again, and have a wonderful weekend!

      Karen

      Reply
  4. Janet Metzger says

    9 May 2026 at 3:17 pm

    Not to be a Debby Downer, but… Each 6-8 item (list) search on chatgpt uses 500 ml water. Factor that cost into your decisions on AI usage as it relates to data centers. Water (in closed loop systems) used in cooling data centers is lost to the planet as it is mixed with an anti-freeze type chemical and cannot be processed in waste treatment. When it is released or leaked (as it will be) into the environment it poisons ground water. See Louisianna’s Cancer Alley… and what soon may be rural and Black communities in Georgia.
    Thanks for posing the question.

    Reply
    • Karen Commins says

      9 May 2026 at 3:24 pm

      Hi, Janet! Thanks so much for the note! I alluded to those dangers in the article, but it’s important to speak precisely about them.

      I actually didn’t know how much water is lost or about the chemical that can’t be processed in waste treatment plants. These environmental and community impacts are much worse than I knew.

      I’m full of admiration for your being such a strong and informed activist! Your information makes the dilemma a no-brainer in choosing not to use generative AI.

      Thanks for your help!

      Karen

      Reply

Have a comment or question?Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Primary Sidebar

Blog and Newsletter Subscriptions

Categories


Blog Archives

Most Popular Posts

  • Public Domain Narration Headquarters
  • Planning Your Trip to Public Domain World
  • How To Become an Audiobook Narrator
  • Audiobook Narrator Self-Assessment Quiz
  • Finding Your Own Road to Tara
  • Embracing Life and Work
  • ACX U - Acting With Intention
  • 4 Keys to Becoming a Successful ACX Audiobook Producer
  • When the Author is 6 Feet Under
  • Links to Help Narrators Research Rights Holders to Books

Links Section

  • NarratorsRoadmap.com
  • AudiobookMarketingTips.com
  • My Tuesday Tips on Twitter
  • Press Page
  • AudioForAuthors.com

More of My Articles

ACX Blog:
  • A Narrator’s Look At Audiobook Marketing Part 1
  • A Narrator’s Look At Audiobook Marketing Part 2
  • How to Act Like An Audiobook Narrator
LinkedIn:
  • Setting Boundaries in Your Voiceover Business
VoiceOverXtra:
  • For a Good First Impression With Audiobook Producers,
    Share Your Audible 'Customer Rating'

Karen@KarenCommins.com

© 1999-2026 Karen Commins // Site design by Voice Actor Websites // Affiliate links to Amazon are used on this site.