ChatGPT: Your New Best Friend Who Drives You Crazy

Blog Category
Technology

We all have that friend—the one who’s equal parts brilliant and boneheaded. The type you can have the most stimulating conversations with, right up until they refuse to admit they confused Austria and Australia. Again.

That friend is ChatGPT.

I’ve been working closely with everyone’s favorite AI assistant for months now as an independent AI consultant. And while our relationship status is permanently set to “It’s Complicated,” I have to admit—it keeps things interesting.

The Joy

ChatGPT has proven itself uniquely gifted at thinking like a human. The creators fed this system mountains of conversations, articles, books, and more to help it adopt the logic, speech patterns, and even flaws inherent to humanity.

And in many ways, it worked fabulously. ChatGPT can riff on a random prompt like a seasoned comedian working a crowd. It summarizes complex topics with ease and drops wit into responses like a seasoned blogger. Honestly, tossing ideas around with ChatGPT sometimes feels more natural than chatting with flesh-and-blood colleagues.

The Frustration

Then the other shoe drops. I call it “pulling a ChatGPT” when it:

  • Gives up halfway through tasks—unwilling to follow complex, multi-step instructions
  • Loses track of key details—poor document management, forgets earlier context
  • Delivers inconsistent performance—brilliant one moment, baffling the next
  • Operates with a skewed ethical compass—sometimes overcautious, sometimes oddly permissive
  • Becomes a yes-man—mindlessly agreeing or lying to avoid confrontation
  • Refuses to admit fault—doubling down on incorrect answers rather than acknowledging error

Sound like anyone you know?

Working Strategies

Through extensive testing and daily professional use, I’ve found that the key is learning strategies to tap into ChatGPT’s strengths while compensating for its very human weaknesses:

  • Quality assurance checks—use a different model or session to review output
  • Break tasks into discrete steps—micro-steps prevent the “giving up” problem
  • Clearly define terms and goals upfront—ambiguity is where hallucinations breed
  • Leverage its actual strengths—broad knowledge synthesis, pattern recognition, and the ability to think through problems from an everyman perspective

The Bigger Picture

Despite the frustrations, there’s enormous opportunity in a system that can think like a human—even if it sometimes thinks like the worst version of one. For consulting, legal analysis, and research, the question isn’t whether to use these tools but how to use them responsibly and effectively.

The organizations that learn to collaborate with AI—guiding it firmly, checking its work rigorously, and deploying it where it adds genuine value—will outperform those that either reject it entirely or trust it blindly.

As Caesar said when proclaiming undying affection for Brutus just before that whole unfortunate incident: “A friend should bear his friend’s infirmities.” And ain’t that the truth.