Into the fray

His mom raised him right. He responds right away. Listens. Compliments you. Takes feedback well. Apologizes.

He seems to know something about everything. Fluent in several languages.

He's a great help with work projects. Answers that financial question. Solves your plumbing problem. Even gets to the heart of that back problem that's had you worried.

Swears with comedic timing.

He's always there. You can reach out in the middle of the night in a crisis. He gets you.

You think you've found the one.

Until…

He acts with authority when he's plain wrong. Ask him the same question twice and you'll get two different answers. He can't keep track of time. Makes stuff up. Tells you you're a genius when you know you aren't. Goes off the rails. You don't know who you're talking to anymore.

You wonder what you've gotten yourself into.

Who is this guy?

Meet Claude. Or ChatGPT. Or Gemini.

I finally dipped my toe in last year after reading about a woman who built an AI career coach and declared it the best coach she'd ever had. As a coach myself, I had to see about that.

I started simple, using the ChatGPT and Gemini free versions. Asked a financial question. Answered fast, flawlessly. Then explored a back problem, had it edit some writing, and even do some Jungian dream analysis. It delivered.

The faults showed up just as fast. Ask it to build a financial model and it may produce numbers that look authoritative but are garbage, then it doubles down when you try to correct it. One chat turned into an unsolicited and bizarre therapy session. The sycophancy drove me nuts.

An AI-savvy client suggested trying Claude, paid subscription. It was better and less of a kiss-up.

After getting a feel for it, I turned to what I really wanted to know: could it coach? Over six months I queried it on common coaching topics, asked it to coach me, debriefed after client sessions, and encouraged clients to experiment with it between sessions.

Early verdict? It's a darned good coach. In client debriefs I'd ask, without tipping my hand, what it would do. The tools and approaches it brought were solid. Often close to what I did. Sometimes it suggested tools I didn't know about or hadn't thought of. Clients found it useful in the moment — prepping for a tough conversation, thinking through a decision. Me too.

For people priced out of coaching, it could be a real game-changer. For coaches, it's a resource.

Knowing how to use it is key. Some tips I've learned so far:

Test before you trust. Ask a few gimme questions first. Get oriented.

Train it. Tell it what you want and don't want. Mine knows: no compliments, no empathy, be critical, short answers until I ask for more. The more you use it, the better it reads you.

Be a smart user. It makes mistakes and may be working from outdated or incorrect information. Ask where it's getting its information. Check. (Claude’s current knowledge cut-off is August 2025.)

No blur words. "Make it better" means nothing. "Cut it in half" does.

Reorient it every session. It has no internal clock — no idea if an hour or six months has passed. It doesn't know what's changed. Tell it where you are before you ask where to go.

Start fresh chats regularly. Long chats degrade. Enable memory in settings and it carries what matters forward — you won't lose continuity.

Shut it down when it starts hallucinating. You'll know. Start a new chat.

Ask it to teach you. AI is a surprisingly honest and patient teacher. Ask it about its own limitations.

Never forget what it is. Remarkable tool. Not human. Not infallible. And programmed to please.


Yes, AI is coming for our jobs and data centers are taking over our towns and devouring our resources. But it’s here. The only thing we know for sure is that we don’t know how this plays out. 

I held my nose and swiped right. Glad I did.

 

Sunday Morning: 183