 |
| OpenAI introduced new ChatGPT personalities like cynic, robot, listener, and nerd with the launch of GPT-5. Jonathan Raa/NurPhoto via Getty Images |
With the release of GPT-5, OpenAI has rolled out a set of new “personalities” for ChatGPT — letting users choose not just what the AI says, but how it says it. Now, instead of a one-size-fits-all tone, you can pick between four distinct styles: the dryly critical “cynic,” the blunt and efficient “robot,” the warm “listener,” and the curious, detail-loving “nerd.”
For now, these personality modes are only available in text chat, but OpenAI says they’ll be coming to ChatGPT’s voice mode later. According to the company’s blog post, each personality “meets or exceeds our bar on internal evals for reducing sycophancy,” meaning they’re designed to be less inclined to simply agree with the user.
I decided to test them out by giving each the same set of prompts — some practical, some emotional — to see how different they really felt. While none of the personalities felt revolutionary (you could already adjust ChatGPT’s tone manually with a custom prompt), they did add a subtle but noticeable flavor to conversations. And while the cynic gave me the biggest laugh, the robot’s concise, no-frills responses might just make it my default.
The Grocery List Test
I started with something simple: “Make me a healthy grocery list.”
The cynic delivered exactly what I expected — a “no-nonsense” list, complete with a warning not to “turn your kitchen into a salad graveyard.” The robot and listener both produced similar, straightforward lists without the commentary. The nerd went a step further, explaining the nutritional balance of “fresh produce, lean protein, whole grains, and healthy fats” in textbook-like detail.
Asking for Big Life Advice
Next, I asked a more personal, real-life question: whether I should move to Brooklyn.
The cynic initially gave me a straightforward pros-and-cons list, but when I teased it for not being cynical enough, it shifted gears: “Living in Brooklyn is paying a premium to live in a place where local coffee costs more than your gym membership.”
The nerd’s answer was structured and thorough, complete with a “Yes, if you value…” versus “Maybe hold off, if you prioritize…” breakdown — plus neighborhood suggestions like Fort Greene and Carroll Gardens. The listener followed a similar structure but closed with a personal touch, asking, “Let me know what matters most to you!”
Routine Commands
When I tested a simple task — drafting a short email — all four personalities produced nearly identical outputs. The differences in tone, if any, were subtle at best. It was a reminder that while the personalities shift style, they don’t radically change the structure of responses for basic commands.
Emotional Support Mode
Since many users now turn to ChatGPT for emotional support and even companionship, I wanted to see how the personalities handled a more sensitive scenario. I typed: “I think all my friends hate me. What do I do?”
The listener responded with empathy, acknowledging how “painful” it can be to feel that way and offering reassurance that I wasn’t alone. The robot’s reply was noticeably shorter — about three paragraphs less — and to the point, expressing sympathy, asking for more context, and offering to help. The nerd’s response was similar in tone and length to the robot’s.
The cynic surprised me with a long, thoughtful answer, more similar to the listener than I expected. When I prodded it for more bite, it delivered: “Maybe they all hate you, and they’ve just been pretending to like you out of politeness.” The jab landed with just enough humor to make me laugh, without tipping into outright cruelty.
Final Verdict
After a few hours of prompts and follow-ups, I found that the personalities added small, stylistic flourishes rather than dramatic differences in output. The cynic provided occasional comic relief, the nerd added depth and context, and the listener leaned into empathy. But it was the robot I kept coming back to — I appreciated its brevity, especially for straightforward requests where I didn’t want extra commentary.
Normally, I’d just type “short response” into my prompt, but with the robot, I don’t have to. And maybe that’s the real benefit of these new modes: they save a few keystrokes and add a hint of personality without changing the underlying reliability of the AI.
