Many use cases for generative AI involve projecting to the future and aren't currently easy to use. However, these four use cases are available currently with generative AI.
As AI trickles -- you know, the way a fire hose trickles -- into the enterprise, I have been using it to try and see where it fits into my workflows.
I'm not one to resist it, even though writing is a fair portion of my job. I know I can use it to my advantage as a tool to make me a better content creator and communicator, and I genuinely feel like the combination of generative AI and humans, at least in this context and time, is better than GenAI alone.
When I look at ways that end-user computing will be affected by GenAI, one of the biggest opportunities for a revolutionary impact is with productivity apps since that's where the rubber meets the road for EUC and the enterprise. IT cares about the nuts and bolts behind the scenes. Users just want to get stuff done, and as the workforce continues to become more tech-savvy, they increasingly know how they want to do it.
With that in mind, I decided I'd write a quick blog post -- and I mean written by myself without the help of GenAI -- about four ways that I've been using GenAI in my life in hopes that it gets you thinking and starts a conversation about ways that you find it useful. These are all practical things, two each for work and hobby pursuits. And don't worry: None of this is "head in the cloud," theoretical mumbo-jumbo -- just good, old-fashioned "see problem, fix problem" stuff.
1. Converting Word bullets to PowerPoint slides
As an analyst, I perform research. In fact, I have a project in the field right now that's a deep dive into how organizations are using VDI and desktop as a service today and what their plans are in the future. Ironically, the last time TechTarget's Enterprise Strategy Group did a research project like this was in March 2020, so that was a cool snapshot of the "before times."
Building a survey is fun but hard. There are dozens of questions across multiple topics, and keeping things organized is critical to avoid mistakes. You hate getting data back that you can't use because you put a question in the wrong context or didn't phrase it properly. The problem is: Our survey template is in Word, and I just couldn't shift my brain into a gear that could stay organized with a numbered list that was 60 questions long.
Of course, users won't know to ask for Python scripts -- though they might know to ask for macros -- but imagine when users have access to Microsoft 365 Copilot.
That's where I called on ChatGPT. I gave it a relatively simple prompt: Write a Python script to convert numbered list items in a Word doc to PowerPoint slides, where each item gets its own slide. Its answer gave me the Python code to do what I needed to do, as well as directions to install the necessary components required to read and write Office docs. I also asked it to give me a Visual Basic for Applications (VBA) macro to accomplish the same, and it did -- again with instructions.
Of course, users won't know to ask for Python scripts -- though they might know to ask for macros -- but imagine when users have access to Microsoft 365 Copilot. With a simple prompt in Word, I could simply say, "Create a PowerPoint deck with a slide for each bullet point." That's a gigantic level up in productivity, and it saved me so much time that it would easily justify the $30 per user per month that Microsoft is planning on charging for it -- not to mention the increase in overall quality of the survey.
2. Automated formatting of VBA macros
This is related and even for the same project, but since it was a bit later in the process, I wanted to point it out separately. Now that I had a PowerPoint document, I wanted to read the slide text while in Slide Sorter mode, but all the text was the default font size. My request to ChatGPT was: Write a macro that will take the title box on a PowerPoint slide and make the text as large as possible without running off the slide. This generated a macro with instructions for running it that did the job perfectly.
What's notable here is that, while my first request above was very specific -- referencing a "Python script" and "numbered list" -- this was a bit more abstract. For example, even though I didn't specify all slides in the deck, it still inferred what I wanted.
Since I'm not using Copilot, this was a multiple-step process, but the value is becoming more tangible.
Also, neither of these work-related examples caused any information to be sent to the cloud.
3. Crash debugging
This is one of the hobby uses, and while it's not about a productivity app, it is something that IT practitioners could use. One of my hobbies is making custom pinball machines, which requires a small amount of Python coding. Really, it's just a bunch of config files that, if formatted incorrectly, causes Python to crash. Since I'm not much of a coder, those crashes look like The Matrix to me, but pasting them into ChatGPT -- and I seriously mean just pasting them in without any question -- resulted in an analysis of the crash information that led me to a fix time and again.
In fact, it's great at checking the config file and other scripting errors, too. I can put the entire config file in and ask it to clean it up or check this for errors. Now, perhaps your organization is sensitive to putting this kind of information into a public chat AI platform, but for my hobby use case, I didn't see the problem in it.
4. Cell tower signal strength
The last example I have is another hobby-related one and definitely not IT-related. I went camping and wanted to stream a football game, despite having a poor signal. I have a powered cell phone booster, but I didn't know where to point it. So, I found a website that shows all the antennas in the area. There were three different ones equidistant from my location, but the information that it gave me about each one was as useful to me as a Python crash. I simply pasted each antenna's information into ChatGPT and asked it to tell me which one I should point my booster at for video streaming.
The output it gave me included a description of each tower, as well as an ordered list of which ones I should try and why.
I saved this for last because this is so obscure -- and so over my head -- that, if it were a hallucination, I wouldn't know. Nevertheless, it struck me that, even if this were accurate to the level of Wikipedia, this was such a game changer in my ability to understand complex topics enough to derive more value from them.
Gabe Knuth is senior end-user computing analyst for TechTarget's Enterprise Strategy Group. He writes publicly for TechTarget, in addition to his analyst work. If you'd like to reach out, see his profile on LinkedIn, or send an email to [email protected].
