I'll admit it: I often annoy my fellow humans. Just like some people prefer aloof cats to Labrador retrievers, some people get exhausted by my extroversion and geeky enthusiasm. And even the people who love me can get aggravated by all the ways I'm impulsive, bossy, overwhelming and judgmental.

Fortunately, I have found a new mentor that can help with what are often called "soft skills"-the interpersonal, touchy-feely stuff that I sometimes get wrong.

Yes, I'm now turning to robots-well, artificial intelligence-as a path to becoming a better human.

Like me, AI can be a bit of a tone-deaf know-it-all, more inclined to dispense advice than to truly listen or empathize. But it also has something I don't: instant access to billions of examples of kindness, tact and emotional intelligence-and none of the ego that makes change so difficult for the rest of us. That combination has made AI a surprisingly effective coach in my quest to become a more thoughtful, compassionate person.

I got my first inkling of this potential when a colleague dismissed my project idea by saying that I lacked the relevant expertise. It doesn't take much to get me riled up, and when a colleague sends me an email that feels condescending or dismissive, my superfast typing speed suddenly becomes a liability, because it only takes me a few minutes to pound out an infuriated response.

I immediately dashed off a "Who do you think I am?" reply. Then I thought to show my colleague's offending email to an AI, which didn't have all my baggage of past gripes with this co-worker or past indignation at being similarly dismissed.

The AI helped me see that I wasn't going to improve the situation by firing off an email that amounted to a highlight reel from my 30-year career; I simply needed to provide a little bit of information that addressed my colleague's specific concerns. Once AI took the chip off my shoulder, I was able to reply warmly and constructively, instead of giving in to impulsive, ego-driven irritation.

Snap judgments

Along with my impulsiveness, AI has helped me resist another bad habit: making snap judgments. Plenty of carbon-based life-forms have urged me to slow down and avoid rushing to conclusions, but to me, even a few seconds can feel like a lifetime.

Take my reaction to the colleague who told me we'd have to change the timeline on a joint project, thanks to a surprise in her schedule. I immediately concluded that this was a sign that she was undependable. How had she committed to this project without knowing her availability? Was she even telling me the truth about her scheduling snafu? Clearly she was either a flake or a liar.

I fired up the AI and let loose a good long tirade about my flaky colleague...and felt as lousy as I always do when I give voice to my most judgmental instincts. Instead of encouraging me to stew in my own mean-spirited juices, the AI asked me a bunch of questions that slowly led me to reflect on what I liked and respected about Little Miss Flaky. The truth was, she was generally well-organized, reliable and extremely proactive; this scheduling glitch was a departure from the habits that had made her a valued collaborator. Pretty soon, I had shaken off my judgy ill temper, and returned to a state of benevolent appreciation for someone who was, for the most part, a pretty great colleague.

Want some advice?

There was still one place where I thought I could count on AI to be as flawed as I am: my bossiness. Like the robots, I respond to almost anything by immediately offering a practical solution.

Apparently, this trait is as annoying in humans as it is in AI. Pretty much everyone in my life has at some point asked me to lay off with the advice, already. But it's soooooo hard when you know the answer! (Claude and ChatGPT, I know you feel me on this one.)

But as I've learned from being on the receiving end of uninvited advice, sometimes you just want someone to listen without immediately telling you what to do. I have used AI to work on this capacity, and to become, well, if not exactly the world's greatest listener, maybe 10% less bossy.

There was the time, for instance, when a friend was mulling over a family challenge, and I thought I had the absolute perfect solution for juggling her personal desires with her caregiving commitments. My solution was so ingenious that I fired up ChatGPT and asked it to lay out all the evidence in support of my plan.

Instead, the AI did what I hadn't, and asked my friend to talk a bit more about her personal circumstances: What were the higher-level goals and values that shaped her decision-making? What were the aspects of her caregiving that provided a sense of meaningful connection, and what were the parts she found confining?

ChatGPT peppered these follow-up questions with a few gentle jabs at my own pressure tactics-enough for me to see that my focus on solutions might actually be stressful or uncomfortable for my friend. It was humbling to see how the AI got my friend to open up in a way that I hadn't, and I recognized that its questions were much more helpful than my answers. Once the AI interrupted my bossiness, I was able to stop pushing my solution, and instead, ask the questions that gave my friend the space to find her own answers.

I can't claim that I have now quit the advice-giving habit. But I have eased up, and I'm now a bit more likely to make people feel empowered rather than overwhelmed.

Maybe it shouldn't take an AI to teach me such fundamental lessons about being a good human: Coach and encourage rather than info-dump and fix; listen and learn instead of rushing to judgment; assume good intent instead of getting defensive and reactive.

But if there's one thing I've recognized from all the time I spend with AIs, it's that it's a lot harder to be a good human than to know what a good human looks like. That's why it takes a ton of work, a ton of feedback and a ton of practice. Instead of wearing out the patience of my friends and colleagues, I'm getting some of that practice with AI-so I can show up at home and work as a better human.

Write to reports@wsj.com