ChatGPT: My brave new world
Shhhh….Don’t tell anyone, but I have a confession to make: I’m playing around.
Get your mind out of the gutter. It’s not what you think.
All I’ve done is invite ChatGPT into my life.
Like many people, when I first heard about this new artificial intelligence thing that could research, write, maybe even think (!), all in a matter of seconds (!!), I was in complete denial. The very idea of “generative AI” was a foggy blur in my mind, and I simply didn’t want to know.
As I read more and more about it, I started to become a little fearful. Maybe even a little angry. (Funny how those two often seem to go together.)
How could this be good or useful? I asked (to no one in particular). Especially when it comes to creativity. There, I agreed completely with singer-songwriter Nick Cave. The idea that something produced by an algorithm could somehow equal an authentic human creative struggle seemed abhorrent. Nein. Nein. Nein.
But then, I consciously decided to get curious. Prompted by Michael Katz, a fellow communications professional, I decided to dip a toe into the water. I created a free account and asked ChatGPT-3 to write a few paragraphs of text. I didn’t exactly love what the “chatbot” spit out, but it was a start. Plus, it did get me thinking about making some changes to a page on my website.
Around the same time, a client asked me to edit a company announcement. I dutifully did my job—real human doing real human work—but on a lark, I decided to feed the original into ChatGPT-3 to see what the chatbot could come up with.
Again, ChatGPT didn’t produce anything earth shatteringly brilliant. (To be honest, it sounded like a reasonably well-written Grade 10 essay. A bit trite, a bit formulaic.) Nonetheless, it gave me an idea for a new line, and that line made it into the edited version I delivered back to my client.
Then I got reckless. I threw caution to the wind and agreed to pony up $20 US every month for a subscription to ChatGPT-4.
Tech geeks will happily explain all the complicated differences between ChatGPT-4 and the free version (ChatGPT-3). But in simple terms, as I understand it, the paid version is faster, more powerful, and more nuanced. It’s also multi-lingual and multi-modal, meaning it can draw on all kinds of source data, including text, images and graphics, to create its output.
Here’s a sampling of what I’ve managed to generate with ChatGPT. Each was done in minutes, if not seconds.
Not too bad, eh?
It’s still early days for ChatGPT, especially for me, but already I can see how the tool, when used effectively, can actually boost productivity and inspire creativity. Those most familiar with ChatGPT will tell you it takes about 10 hours of playing around to become proficient at getting what you want.
To help you get started, here are three things I’ve learned about ChatGPT so far:
1. ChatGPT is different from Google.
Recommended by LinkedIn
Google searches the internet to provide relevant results. It’s a one-way broadcast. You put in a search term, you get a list of results. Rinse and repeat, if desired.
In contrast, ChatGPT is an example of “generative A.I.”, meaning it mimics human conversation. And, like the best conversations, the more you interact, the better the outcome. The first result may provide broad strokes, but the deeper you go—the more willing you are to pause, consider, digest, and ask follow-up questions—the more satisfying the whole experience will be. Just like “real” conversation.
2. Prompting is key.
In ChatGPT world, you don’t enter search terms, you “prompt.” While prompt is effectively another word for the query, question or instructions you give ChatGPT, there is a bit of an art to issuing good prompts. And, like most things in life, the more you practice, the better your results.
What makes a good prompt? Tips I’ve gleaned include:
Hmmm….sound familiar? Plain language, anyone?
If you’re interested in learning more—from a “Professional Prompt Engineer” no less—check out Anna Bernstein’s helpful little video, “6 Tips to Writing Better Prompts”.
3. Think of ChatGPT as an intern.
As Ethan Mollick, Professor at the Wharton School of the University of Pennsylvania, has written, “AI is not good software. It is pretty good people.”
Mollick makes it clear that ChatGPT is not a substitute for humans. But it does excel at doing human tasks—like chatting, writing, analyzing and coding. (I can't profess to know anything about that last one.)
As Mollick says, “An AI may not be error free, but [it] can save you lots of work by providing a first pass at an annoying task. You can even teach it to do better by providing examples of good output.”
Examples of what ChatGPT is good at:
Beware that ChatGPT can “hallucinate” (a.k.a. lie or make stuff up), so remember that you’re responsible for checking accuracy. As you would with the work of any junior. But, as the old saying goes, nothing ventured, nothing gained.
Remember this: It’s worth taking the time to experiment with ChatGPT—or another generative A.I. of your choice. Think of the chatbot as a pretty good intern, one you can prompt to helpfully get the ball rolling on all sorts of tasks.
This post originally appeared in The Clarity Chronicles newsletter (January 2024).