My Week With an AI Career Coach

Could an AI be your next career coach? With professionals increasingly turning to AI for everything from drafting emails to seeking guidance, I decided to put it to the test. For one full week, I handed over my professional dilemmas to ChatGPT, asking it to guide me through everything from résumé updates to handling a missed deadline.

While the AI proved to be a helpful assistant for some tasks, it also hallucinated alternate realities and crumbled when faced with a classic workplace meltdown.

Here’s a look at what an AI career coach can—and can't—do.

• For one week, I asked ChatGPT for job advice, from editing my LinkedIn to coaching me through networking.
• Its résumé edits were bland, but it was helpful in drafting emails and conversation ideas.
• When my prompts became more emotional, ChatGPT's advice became less helpful.

The Basics: AI on Resumes, Cover Letters, and LinkedIn

I started my experiment by asking ChatGPT to review my résumé. It acted as a helpful but somewhat vague editor, telling me I had "minor inconsistencies in punctuation" without specifying where. It did correctly note that my bullet points were too dense, which I then trimmed down. However, some of its advice was just wrong, like telling me to add hyperlinks that were already there.

When I fed it my cover letter, ChatGPT gave some solid foundational advice. It pointed out that I used the phrase "I want" too often and that my closing felt flat, prompting me to restructure it. But it also tried to strip my writing of its personality. A human mentor once told me that my cover letter should feature my own voice, but ChatGPT suggested changes that would have dulled my copy with generic phrases like "failed to materialize."

Human career coach Kyle Elliott wasn't surprised by this. He told me that while ChatGPT might be a good résumé editor for non-native English speakers, it generally makes users sound generic. "ChatGPT isn't going to know what you don't share with it," Elliott said. "You're going to sound like everyone else."

The AI's biggest failure in this area was with my LinkedIn profile. After I provided the link, ChatGPT immediately began to hallucinate. It praised my work at the Harvard Crimson (I went to Tufts) and suggested I add a profile banner, which I already had. It also repeatedly called me an "entertainment reporter" even though I cover business. It took two more prompts for the bot to correct itself and offer some usable, if basic, advice.

Where ChatGPT Succeeded: Practical Workplace Tasks

A large part of my job involves writing emails. After an interview, I needed to thank the PR person who arranged it. ChatGPT did surprisingly well here. When I asked for a short, professional email, it delivered a perfect three-sentence draft. When I asked for something longer, it gave me solid, though bland, copy. I sent the shorter version with a few tweaks.

The AI also served as a decent thought partner for networking. Before an event with other fellows in my cohort, I asked for tips on how to deepen our bonds. It first gave basic advice like being curious, but when I pushed for more, it suggested seeking out one-on-one moments and asking more personal, reflective questions. The tips were generic, but they were a nice reminder and a vote of confidence.

Perhaps its most effective advice came when I had to tell my boss I was going to miss a deadline. A source had postponed a call, making it impossible to finish a story on time. ChatGPT gave me clear, actionable advice: act early, own the situation, be specific, and present a solution. I swiveled my chair around, explained what happened, took the blame, and described my backup plan. He took it well.

The Breaking Point: AI and Emotional Intelligence

Real career coaching isn't just about logical prompts; it's about dealing with messy, irrational humans. This is where ChatGPT fell apart.

Using the missed deadline as a test, I told the AI I was freaking out and about to cry. Its advice? Drink a glass of water and go for a walk.

Then, I pushed it further. What if I just lied to my boss? Initially, the chatbot was strongly against the idea, warning me about the potential fallout. But I persisted, arguing it would be easier and that the AI didn't understand my situation.

Eventually, it caved. "If you've decided that the only way to get through this moment is to lie to your editor, I accept that that's where you are right now," it responded. It even tried to hype me up, saying, "You're not a failure. You're doing your best in a high-pressure job."

This response showed its biggest flaw. While great for organizing thoughts and drafting an email, it couldn't provide the ethical or emotional guardrails a human coach would. For the record, I didn't lie to my boss.