I’ve done a pretty good job reading up on AI.
I know, for example, what The Atlantic and The New York Times think about the future of the college essay.
And I played around some with ChatGPT. Most memorable was the essay my son had it write. The output was an 80% approximation of the college essay he’d written last year, terrifyingly similar in structure and tone. I also tried to get it to write a post for this blog. The results were watered-down and uninteresting.
I also made sure to talk to my team members about how they were using it to save massive amounts of time coding qualitative responses to questions—at 60 Decibels, we speak to hundreds of thousands of people each year, and turning their open-ended qualitative responses into quantitative data is a core part of our business model.
But I hadn’t used it to solve any meaningful business problem that I, directly was working on.
Until last week when a team member described a thing she had done, and I decided to do it too.
She kindly outlined the steps and did a short Loom video to explain it.
Then I mucked around some, adjusted what she did, and worked with the output ChatGPT gave me.
The gap between what I thought the tool could do and what it actually does (and does not do) was pretty big. And I’ve just used it once—I’m positive I’m just at the beginning.
If you’re like me—if you haven’t done any real work using ChatGPT or another LMM tool—I’d encourage you to take that next step: find something real that you need to get done, and figure out / have someone help you figure out how to do it.
You’ll learn a lot, and you’ll also start noticing more and more situations where AI might be helpful to you.
Until now, I thought of this as a tool that was out there.
And I wasn’t actively thinking about how I could use it when a new task came along.
That’s a recipe for falling behind if there ever was one.