Happy Friday!

Vulnerable collaboration

For almost 20 years, I’ve worked with a writing collaborator. David and I have penned three musicals and several screenplays, and we’re both currently at work on novels. (The novels aren’t co-written, but it’s nice to have collaborative feedback!)

As collaborators, we have to be completely comfortable saying anything to one another—sharing half-formed ideas, giving constructive feedback. The best collaborators allow you to be vulnerable, because you know they will never intentionally hurt you.

When you work together as long as we have, and you’ve been as open with each other as we have, you begin to know how the other person thinks. I have a model of David’s thinking in my mind, and David has a model of my thinking in his mind.

Artificial collaboration

We now have a tool—all of us—that can serve as an artificial collaborator. ChatGPT goes well beyond being a “sounding board” and into the realm of “creative partner.” But just like collaborating with a real person, you have to bring high-quality work and thinking in order for the collaboration to produce usable fruit.

If I consistently bring garbage into my writing collaboration with David, he would remain my friend, but he would rightly end our collaboration. In contrast, ChatGPT would happily engage with my garbage and spit its own garbage right back at me. That’s not a healthy collaboration.

But questions and half-formed ideas are not inherently garbage. And, just like with a (good) human collaborator, you can bring half-formed ideas and ask questions. You can be vulnerable—what’s ChatGPT going to do, laugh at you?

A future may arrive when AI knows you so well that it can do all of your thinking or writing for you. But that time has not yet come. Which means…

💡 Collaborate with AI, but don’t rely on it 💡

👆 That’s your weekend upgrade.

Quick sidebar. I mentioned that David and I write musicals. I’m the composer/lyricist. And when I’m writing lyrics, I use a tool: a rhyming dictionary.

To write lyrics, my first step is determining the dramatic function of the song in the show. What character(s) will sing it? What do they want? Do they get it? What do they discover?

Once I know the character and dramatic arc, I brainstorm key words and expressions that convey the character and drama. Then I head over to rhymezone.com and find words that rhyme with those key words and expressions.

Now that I have lists of rhymes, I discover connections and ideas to tell the song’s story that I would never have found without prompting. The resulting lyrics are undeniably my own, but the rhyming dictionary helps me explore territory I would have otherwise missed.

Three tips for AI collaboration

To get the most value out of an AI conversation—without sacrificing your voice—use AI like a “rhyming dictionary collaborator.”

(1) Start your ChatGPT conversation only after you know what your purpose is

If I opened a rhyming dictionary and idly searched for rhymes, I might write a song… but it would suck. I need to know what I’m looking for.

Whether you’re using ChatGPT to generate sample task lists or outline blog posts, asking it generic questions will provide generic results. But if you know what you’re trying to accomplish, you can ask better initial questions and follow-ups to get exactly the results you need.

(2) Be you at “both ends” of the process

The ideas for lyrics are mine, and the execution of the lyrics are mine. The rhyming dictionary provides prompts to help me get from idea to execution.

ChatGPT might give you a blog post outline, but if you don’t start with an idea that interests you and finish with concrete writing that comes from your experience, you’ll have Zeitgeist-fueled drivel. You are what makes your work interesting.

When you’re having an AI conversation, be yourself—just like you would be with a real-person collaborator. Follow ideas that interest you. Ask questions you genuinely care about the answers to.

(3) Lean into AI’s strengths

It would be silly for me to strain my brain cycling through possible rhymes for important lyrics. A rhyming dictionary is a much better tool.

Likewise, AI is a lot better than me at, say, immediately churning out a database design. Part of my AP Productivity cohort is building workflows in Tana. If I want to build a simple Recipe & Meal Planning workflow in Tana, ChatGPT is a great place to start. I could say:

Then from the answer, I’d disregard “database-y” stuff like ID fields (except as models for relationships), translate “tables” into “supertags,” and prototype the workflow in Tana. ChatGPT wouldn’t create the entire workflow, but it would help me conceptualize what I need to build.

Another example from my own ChatGPT use: I coach bank tellers to have better conversations and build customer relationships. As part of that, I bring in actors to run scenarios that target specific communication skills. If I feed my relevant training content into ChatGPT, it can generate simple scenarios for those roleplaying trainings. All I need to do at the end is make sure the scenarios will accomplish what I need. That’s using ChatGPT for a job that it will do much faster—and probably much better—than I would.

There’s no “I” in “AI”

When I write with David, we’re a team. But contrary to the popular expression, there is an “I” in “team.” If I don’t bring my perspective, my experience, my interests to the collaboration—in an open, vulnerable way—our team will not succeed. I shouldn’t bring my ego, but I must bring my self.

Working with AI makes this far more essential, because I’m the only “I” present. There’s no specific perspective, no unique experience, and definitely no personal interests coming from ChatGPT’s side. All that must come from you, if you want to make the most productive and valuable use of the tool.

The more you bring yourself to that AI collaboration, the better you’ll understand how best to use the tool. And, over time—just as David and I have “models of each other’s thinking” in our minds—your use of AI will become more and more an extension of your own perspective.

What do I do next?

(1) Take 2 minutes and answer this question: What’s one thing I learned in this newsletter that I can put into practice right away?

By committing to a specific action, you make it much more likely you’ll do it.

(2) Choose one workflow in your life that could benefit from having an AI collaborator.

Where would ChatGPT fit in that workflow? What would you bring to the table? What questions would you ask? And what would you do with the answers?

If this was valuable for you:

Share the newsletter with someone you think would also get value from it! https://rjn.st/weekend-upgrade-newsletters​

Until next time, friends:

Be a great collaborator—even if you’re the only person involved!

R.J.
​rjn.st/links​