Discover more from ECOSYSTEM by JD
5 ways AI improved my work last week.
The AI conversation is chock full of “cans” and “shoulds” right now.
To show you what I mean, here are excerpts from three articles featured in my AI @ Work Magazine:
“AI can be leveraged to provide more personalized learning experiences.”
“Generative AI has the potential to increase US labor productivity by 0.5 to 0.9 percentage points annually through 2030 …”
“A new crop of artificial intelligence tools carries the promise of streamlining tasks, improving efficiency and boosting productivity in the workplace.”
AI is altering our relationship with technology. This is clear. But we still don’t know exactly HOW things are going to change. AI is far from new, but we’re still early in this next wave of digital transformation (buzzword!) fostered by generative AI.
Over time, AI-enabled technology will fundamentally shift how work gets done. Workflows will evolve. Jobs will transform. Productivity will explode. Creativity will flourish. That’s where we’re headed, and we must work together to shape the new version of work. But that doesn’t mean we must wait for the final picture to be revealed. There are immediate, low-risk gains to be had from this tech.
Here are 5 simple ways I used AI last week to improve my work. SPOILER: AI saved me 11 hours!
I partnered with TrainingPros to host an UNwebinar on AI in L&D a few weeks ago. The webinar was “un” because I didn’t bring any slides or deliver a book report on the evolution of technology and the industrial revolution(s). Instead, I reserved most of the hour for attendee questions, but I still ran out of time before I could get to every submission. Every question deserves an answer, so I always commit to respond to unanswered questions via video after my sessions.
This video came in at 1 hour, 28 minutes. I cover all 29 questions submitted during the session. As YouTubers know, the thumbnail makes the video, so I had to design a custom image before I could upload my Q&A. Typically, I go hunting on stock image websites to find assets that fit the session theme. Then, I figure out how to merge my photo with the assets to create the final image.
Since the session was about AI, it made sense to use AI to generate the thumbnail. I logged into Bing Image Creator and entered the prompt “a variety of robots looking into a white circle. “ This was one of the four options generated in under 10 seconds.
It took a few extra minutes to Photoshop myself into the blank space, add a few filters and overlay text to create the final thumbnail.
AI enabled me to generate a custom image that fit my vision in a single attempt. Half of the images I use in presentations are now created using generative AI. This includes my upcoming Modern Learning Mindset workshop with L&D Philly. After all, where else can you find artwork of a Delorean speeding through Center City Philadelphia?
Time Saved: 2 hours
I’m bad with titles. I’m incapable of coming up with short, eye-catching, informative subject lines for articles, emails and webinars. That’s why ECOSYSTEM editions have one-word titles. It’s less a creative choice and more a demonstration of my limited imagination. Plus, I despise clickbait and prefer the “just tell people what it’s about” method. Unfortunately, this doesn’t typically fly with publishers, marketers and event organizers.
I enjoy writing - finding new ways to share ideas and tell stories. I just don’t like writing the titles for those stories. This is especially true for conference sessions. It’s not uncommon for me to spend hours staring at a finished proposal, trying to conjure a 10-word title that will capture people’s imaginations.
I wrote four conference proposals last week, so I had to come up with four titles. Rather than choke on my own word salad, I turned to ChatGPT for assistance. I uploaded the descriptions for each session along with requirements for an effective title (10 words or less, engaging, fun, action-oriented, relatable to an L&D audience). ChatGPT returned 10 ideas for each session title. They were all bad. For example, here’s the list for a session on manager enablement:
"Mastering Manager Magic: Empowering Teams to Excel"
"Supercharge Your Managers: Boosting Team Brilliance"
"Unleash Manager Superpowers: Uniting Growth and Fun"
"Rockstar Managers Unplugged: Igniting Learning Sparks"
"Manager Mavericks: Driving Team Development Dynamite"
"Enabling Managers, Unleashing Potential: Playful Pathways Ahead"
"Rethinking Manager Mojo: Sparking Learning on the Job"
"Adventures in Manager Enablement: A Team-Building Odyssey"
"Unlocking Manager Marvels: Fun-Fueled Leadership Learning"
"Manager Enablement Carnival: A Thrilling Ride to Success"
Would you attend any of those sessions? I would not.
ChatGPT didn’t generate the perfect title, but it did spark an idea. I zeroed in on the word “unleash” and combined it with my thesis statement to arrive at “Managers Unleashed! Enabling the Most Important People in L&D” (coming to a conference near you in 2024).
Time Saved: 1.5 hours
Thanks for reading ECOSYSTEM by JD! Subscribe for free to receive new posts and support my work.
I sign LOTS of greeting cards nowadays. This wasn’t always the case. As a remote employee, I was typically excluded from this office tradition. Someone else would add my name when necessary. Then, my company went hybrid and adopted digital greeting cards. My role is cross-functional, which means I now sign cards for multiple teams.
This wouldn’t be a problem if I signed cards like the average person. For example, what do your write in a typical birthday card? “Happy Birthday,” right? I can’t do that. It just feels wrong to write the same thing over and over again. If I’m going to sign the card, I’m going to SIGN THE CARD. But coming up with 3 or 4 special messages every week takes time. At least … it used to take time.
ChatGPT now signs all of my cards for me. I prompt the bot with insights about the card recipient and their special occasion. Of course, I’m careful not to provide any personally identifiable information, including their real name, company, title, location, age, etc. ChatGPT then drafts a one-of-a-kind greeting card message.
Copy, paste, tweak, done. Bonus: every card now includes the signature “JD (with AI help)” because transparency is important, even with greeting cards.
Time Saved: 30 minutes
I consume EVERYTHING about my field. Newsletters, webinars, podcasts, books, reports. I then curate high-value content on relevant EX themes and share it with my team via Slack. This way, everyone can find useful information without sifting through the noise.
A few months ago, a team member Slacked me with a question about my curation efforts. They were wondering if I could add brief summaries to my Slack messages. They appreciated my work, but they didn’t have time to read everything I shared. A summary could help them determine if an item was relevant to their role. Unfortunately, I didn’t have capacity to write summaries for everything I shared. There was just too much information moving through the pipe for any person to handle.
Enter AI! We applied the same LLM technology used to generate assessment questions to summarize information shared in Slack. The bot reads the full content - articles, PDFs, video transcripts, etc. - and generates a five-point summary in seconds. For example, here’s the summary of RedThread Research’s new (members only), 30-page learning tech report.
It doesn’t provide every detail, but it gives you enough context to determine if it’s worth your time to review the full content (and it totally is in the case of RedThread’s new report!). It would’ve taken me hours to write a summary for every curated item I shared last week. Without AI, it just wouldn’t have been feasible.
Time Saved: 4 hours
Cara and I will be playing Agree to Disagree! - a debate game popularized on YouTube by celebrities like Matt Damon and Emily Blunt. Each round starts with a narrator reading a statement about an L&D topic. Cara and I must then decide if we agree or disagree with the statement and support our opinions.
I pre-record the narration to limit production complexity during the livestream. Traditionally, this means I have to …
Write the narrator script.
Find a coworker with a voice that stands out from everyone on the show that week who is willing to participate.
Send them the script.
Schedule a Zoom recording session.
Coach the voice-over performer during the recording session.
Edit multiple takes together to create the final audio clips.
This process requires hours of work from multiple team members for a 15-minute show segment. So I turned to ElevenLabs to accelerate the process for Cara’s upcoming episode.
Write the narrator script.
Select AI avatar (Victoria).
I generated the nine audio clips needed for the show in five minutes. Tune in to ITK on Wednesday, August 9 at 1130am ET to hear the final product!
Time Saved: 3 hours
Until next time, be well. JD