ChatGPT Study Mode Helps With Decision Making; Led Me to Stay Car-Free

For months, I’ve been paralyzed with indecision every time I walk past a car dealership.

It all started earlier this summer when my partner bought his first car — a five-year-old black Ford Fusion — ahead of a move to Indiana for his Ph.D. program.

Since then, we’ve both been using the car we named Raven, and I’ve become used to picking up groceries with ease and leaving home 10 minutes before a dinner party starts. But with him leaving in three weeks and taking Raven along for the ride, I’ve become overwhelmed with confusion every time I ask myself: Should I buy a car?

So I asked ChatGPT.

And not just any ChatGPT: I chose study mode, a new version that has freshly appeared in the toolbar as a book icon. OpenAI says it “helps you work through problems step by step instead of just getting an answer.” It can also quiz its users and prompt them to explain their reasoning.

I did not have high hopes, having witnessed many disastrous ChatGPT-generated essays that friends in academia have had to grade. Still, the study mode asked me enough well-rounded questions to help me make the unexpected decision to remain car-free.

What I knew before turning to ChatGPT

I live in a co-housing community with a garage to store and charge cars. I also happen to live downtown in a Bay Area city, two blocks away from a Chinatown, where I can find just about any food I need.

Still, having a car would mean not having to frequently turn to Instacart because shopping often overwhelms me. It would mean being able to access our regional park full of redwoods, where there is no cell signal and no chance to Uber back home.

I prefer EVs because they emit less, and I like the peace of mind of knowing that the price of oil, which fluctuates with geopolitics, won’t affect my budget as much.

Most of my friends own cars and started sending me their hot takes. I started watching Instagram reels on car recommendations, until they had fully taken over my “For You” page.

Based on those criteria, I found options like a used Nissan Leaf and a pre-owned Tesla.

A used 2017 Nissan Leaf would only have between 50 and 80 miles of range per charge, but it could cost as little as $7,000, not including taxes. Teslas are a more expensive option, but they do have decent range and technologies that compensate for my lack of skills, plus many people are looking to sell.

In my head, I felt like I spent plenty on rides and deliveries to match the convenience a car could provide.

Fortunately, ChatGPT study mode explained to me that not only was I wrong, but there was so much more to consider.

Study mode asked me surprising questions and taught me new concepts

Study mode started by asking me what level of studies I’m at, to which I explained that I have already gone through grad school, but have other pressing life problems.

I then gave a general description of my circumstances and asked whether I should buy a car.

The AI commended me for making a “thoughtful, not impulsive” decision and explained the concept of being “car poor,” meaning buying a fancier car than necessary or having monthly payments cut into other life expenses.

Since I mentioned grocery deliveries and Uber trips, study mode then prompted me to think about how many times I use these services a week. I get grocery deliveries about three times a month, plus about two Uber trips and two takeout deliveries a week.

Study mode soon crunched the numbers for me. The conveniences I see as indulgences actually cost me around $3,000 a year, but a car would cost me between $6,000 and $8,000 a year, not including the down payment I would need to put down for the car.

“Are the extra ~$3,000 — $4,000 per year worth the added freedom and independence? Do you feel anxious or limited without a car?” the AI asked.

In bullet points, it asked me about factors I hadn’t really thought about before, such as whether I like to go out often, if I enjoy driving, and if I have family who lives far away. It also asked me if I wanted to see a cost breakdown of whether it would be worth it to live further from downtown to lower housing costs, but own a car instead.

As an introvert with no family in this country who mostly spends her weekends with her cat and her next craft project, an answer was beginning to emerge.

But I pushed study mode further by asking about the benefits of an EV and if it would actually save me money. The AI gave me a cost breakdown that compared a Chevy Bolt to a fully gas-powered Toyota Corolla, and while the former obviously emits less, it gives me only a marginal amount of savings.

Based on my own investigation, the cost difference between the two cars appears accurate, but the AI vastly undercounts the cost of insuring both types of vehicles by more than 50%. A quote on Geico for a 2022 Chevy Bolt and a Toyota Corolla made the same year would both amount to more than $450 a month in insurance.

As alternatives to an EV, the AI asked me to consider if carpooling with friends and renting cars when I truly need them would be better options for my wallet and for the environment.

It also said buying a three-year-old car is optimal because that is when depreciation slows down and before maintenance costs start to rise.

Study mode said the bottom line was that if I really, really, still wanted a car, it would suggest a 2022 Chevy Bolt.

Still, I think I’ll pass, because to answer one of the AI’s previous questions: no, I don’t actually enjoy driving. I will save myself the fear of freeways and the panic of not being able to parallel park, and continue my car-free life.

My next question to ChatGPT will be what kind of bike I should get.

OpenAI did not immediately respond to a request for comment.