I wrote this article with no help from ChatGPT

I wrote this article with no help from ChatGPT. That sentence would have made very little sense six months ago.

In the first days after ChatGPT’s quiet release, I listened to podcast episodes and watched videos of people describing how ChatGPT could “write a sonnet about professional basketball in the style of a prohibition-era gangster.” Everyone wanted to see what sorts of tricks it could do, like some sort of curiosity at the county fair.?

It took everyone a little while to recognize just how powerfully productive ChatGPT could be. I found myself down YouTube rabbit holes focused on mastering “prompt engineering" and on automating workflows using APIs, about integrating ChatGPT into Google Sheets to auto-generate hundreds of emails at once.

For myself, though, I've found a happy medium between treating ChatGPT as a quirkly wunderkind and part of a fine-tuned, meticulously engineered productivity machine.

I use ChatGPT as an occasional helper and as an intellectual equal. It’s less creative and less nuanced than I am. It’s also infinitely faster and impeccable at structuring information. If I ask it to explain a topic that I know about, I would recognize its outputs as surface level. But for a new topic, surface level is a great starting point.?

I’ve asked ChatGPT to be my coach (“I want you to act as a Master Certified Coach”), asking me a series of open-ended questions to help me re-start a daily meditation habit. It helped me brainstorm ideas for a father-son golf trip for my dad’s birthday (“It's my dad's birthday, and I want to go on a weekend trip with him. I'm based in Washington, DC, and he's based in Boston.”) In my latest coup, I gave ChatGPT a detailed powerpoint slide and asked it to recreate 7 others following the same format, and it performed beautifully.?

Over the weeks, I've gotten a lot better at helping ChatGPT help me. Like with any tool, I know I need to be clear, in my own mind, about what I'm trying to accomplish. That can be in how I craft my initial prompt. But it can also be after I read ChatGPT’s response and saying “that’s not what I’m looking for. Could you make it more X, Y, or Z?” It follows the same principles of giving clear feedback to a person—I have to be specific, and I have to know what "good" looks like.

Ultimately, ChatGPT has its own quirks and logic, and I've only only gotten to understand know how it works by testing it, seeing where it wows me and where it lets me down. If I'm asked how to best use ChatGPT I'd say, "You kindof just have to play around with it and enjoy the journey, if that's your kind of thing."

I certainly acknowledge the weirdness and anxiety of all of this, about how--as a society--we've opened Pandora's Box and await the consequences.

My biggest concern right now, though, is a selfish one: that my own skills could atrophy. It’s admittedly harder for me now to stare at a blank page, because I no longer have to. It’s like driving a car without a backup camera or blind spot detection—I love those new features, but it’s unsettling to drive without them.

For now, though, I have to say: like when we brought home that first Macintosh computer. I'm having fun playing with this new toy and exploring its possibilities.