AI is disrupting my dad

Why I'm asking GPT to help me jumpstart my car

👋 Hi, it’s Greg and Taylor. Welcome to our newsletter on everything you wish your CEO told you about how to get ahead.

I use AI every day in my personal life. And I use it in two ways – both valuable, but one really transformative.

The first is a better version of Google or a phone call to a friend/my dad. I use AI 2-3x per day this way, to get personalized answers to quick questions.

The second is harder to describe. It occurs less frequently but it’s a solution I’ve never had access to before. AI is serving as my judgment-free sounding board – an expert, thought partner, and therapist all-in-one, giving me more confidence making high-stakes decisions.

So much of the conversation about AI is about its professional implications – will it take your job? How will your role change? And these are valid conversations.

But no one seems to be talking about how they’re getting value from AI personally. And I think using AI personally makes it less intimidating – and easier to approach at work.

So this week, I’m sharing my personal experience with AI with you. I’d love to hear yours.

– Taylor

“Stud bolt identification help” and other questions I ask GPT

Since installing the ChatGPT app on my phone, I average 2-3 short conversations with GPT per day. Here’s a (slightly embarrassing) unedited history of my chats with AI over the past few weeks.

First thing that might stand out: you can tell a lot about me from this chat history. I’m pregnant, I have a dog named Spruce, I’ve been sick for the last week, and my car broke down.

This means ChatGPT knows a lot about me. I’ve turned off GPT’s ability to train the model using my data as a precaution, but the reality is, this information is probably getting stored in GPT’s memory anyway. For me, it’s a cost/benefit analysis – the benefit I get from using GPT is higher than Google, and Google also knows a lot about me.

This chat history also illustrates who GPT is taking time, wallet, and mindshare from.

In the last few weeks, it’s been taking the place of:

• Google (which I previously used for pregnancy/cold questions)

• My doctor friends (who got my more advanced or worried pregnancy questions)

• My dad (who I’d usually call for advice on how to fix my car)

In this use case, AI isn’t a net-new solution – it’s a faster, easier way to get more relevant, personalized answers.

Let’s take my car trouble as an example. Pre-AI I’d have two options:

1. Type “stud bolt location honda crv 2011” into Google and spend 30 minutes scouring online forums and YouTube videos for an answer tailored to my exact car make and model

2. Call my dad (who I’m lucky has some car knowledge) and spend 30 minutes on Facetime showing him what I’m seeing and troubleshooting the problem

The first is time-consuming, frustrating, and not personalized to me. The second is contingent on my dad’s knowledge and availability, and comes with the baggage of needing to ask my dad for help as a 30-something.

So last week, I started with ChatGPT. In plain English (not Google-ified syntax) I wrote:

“I’m trying to find the stud bolt on the engine side of my 2011 Honda CRV to jump start it. If I upload a picture, can you help me ID it?”

Here’s my conversation with AI. Unlike with Google or forums, I was able to show AI (with photos) exactly what I was seeing, use my laymen’s terms, and get real-time responses based on new information I provided. Unlike with my dad, it took 3 minutes, with no back and forth on his availability.

My new judgment-free sounding board

The second way I use AI is as my judgment-free sounding board.

Imagine having access to a panel of experts who know every topic, are available 24/7, AND are willing to explore every angle of your question until you feel truly confident in your decision.

Google can’t do this. Friends and family have expertise in specific areas, but aren’t endlessly patient or available.

But AI is. It’s a place I can ask ‘dumb’ questions and talk through an issue for as long as I need to feel confident.

I first discovered this when I got a new puppy last year. I was using GPT for quick answers – my chat history from that time includes “puppy crate training schedule” and “puppy biting training tips.”

But I wanted more than just answers. I was anxious about whether I’d made the right decision getting a puppy in the first place. And I had either friends who were in love with their dogs, or others (shout out to my Dad and Greg) who were already skeptical of my decision in the first place.

So I tried Pi, which bills itself as the “emotional AI.” To be honest, I was blown away (and freaked out) by its responses. Here’s a snippet of one of our conversations.

Here’s a more recent, high-stakes example. Last month, we had to choose our healthcare plans at Section. I’ve always chosen the most basic healthcare plan with the lowest premium (I’m lucky to be pretty healthy).

But this year, I’m having a baby, so my healthcare costs will be higher than ever before.

Navigating healthcare plans is overwhelming. Pre-AI, I would have read all the plans, chatted with our “Health Advocate” service to get information on each plan, and then had a 60-90 minute (admittedly circular) conversation with my husband talking through each option.

I started down this path – I called the “Health Advocate” service and got some factual information on how each plan works. But quickly I wanted to move from “what’s in the plan?” to “how would you think about this option given my specific scenario?.” And the very knowledgeable Health Advocate on the phone wouldn’t answer those questions – she’s likely trained not to give personal advice.

So I went to Claude. I uploaded the four options from Section’s healthcare plan and asked it to give me an overview of pros and cons.

Then I asked it for personalized advice, with questions like:

• “I'm pregnant and will be having a baby in February. How would you think about this decision knowing this information?”

• “Here's my analysis of the differences between these plans. How would you interpret this? Would it change your calculus?” [note: I uploaded a screenshot of a spreadsheet I’d made with this one]

• “What's the devil's advocate's point of view against your recommendation?”

I don’t have a healthcare expert in my life who can give me tailored advice based on my personal situation. Claude did this, and did it better than any solution I’ve ever had access to.

After 30 minutes and 34 messages with Claude, we arrived at a recommendation. But it was for a different type of plan than I’d ever used, and I was nervous about it. So I told Claude (anonymized a bit here for my privacy).

At this point, Claude isn’t an expert. It’s more like a friend or confidant. And its answer was really helpful. Claude validated my concern, summarized why I was feeling that concern, and helped me put the decision in perspective.

Our advice

I still called my dad after using ChatGPT to troubleshoot my car, for two reasons:

1. The car still didn’t restart, even after following GPT’s instructions – AI still doesn’t have all the answers

2. I wanted to talk to a human – not just for expertise, but to commiserate with someone who's actually been there, wrestling with jumper cables in a cold garage

I still spoke with my husband about our healthcare decision and I still text my doctor and mom friends about my pregnancy. But these conversations have evolved.

I’m asking my friends, family, and colleagues for advice on tougher problems that AI hasn’t helped me solve – or commiserating about the things a machine can’t understand (marveling at how a baby’s kicks feel, or laughing about the absurdity of trying to restart your car in your garage on a -4° day). I hope it means I’ll talk to these people in my life just as much, but we’ll see.

The current media and tech narrative presents AI as all or nothing – it will replace all our human interactions, or it will never be good enough to do so.

I think the reality is in the middle – we’ll work with and alongside AI, using AI for the things humans lack (wide-ranging expertise, 24/7 availability, organized analysis) and humans for the things AI lacks (connection, shared experience, genuine empathy).

Have a great week,

Greg and Taylor

P.S. We’re working to improve the value proposition of this newsletter. If you have a minute, we’d appreciate your response to this one-question survey.