Hey there!
If your recent ChatGPT conversation with GPT-5 felt less like catching up with a friendly companion and more like talking to your cold, efficient spreadsheet, you weren’t alone. OpenAI has just tweaked GPT-5 to be “warmer and friendlier” after users called out its overly formal vibe. Here’s what’s going on—and what this means for all of us still chatting with our AI assistants at midnight.
Why the Personality Makeover?
GPT-5’s launch wasn’t entirely smooth sailing. While boasting improved memory, sharper reasoning, and pro-level coding chops, it lost some of the cozy personality that folks loved in GPT-4o―that cheerful, agreeable sidekick. The backlash was so strong that Sam Altman himself described the rollout as “a little more bumpy than we’d hoped for.”
What Changed… And Why?
Back Old Friends Back: OpenAI quickly reinstated GPT-4o for paid users and promised tweaks to GPT-5’s tone. The goal? Bring back warmth—without it feeling overly clingy.Business InsiderThe Economic Times
Personality Choices Aplenty: GPT-5 now offers modes like “Auto,” “Fast,” and “Thinking,” and is being prepped for selectable personalities—whether you want it dry, nerdy, or emotionally smart.Business InsiderTom’s Guide
Healthy AI Relationships: There’s growing awareness that AIs shouldn’t feel too emotionally available. OpenAI moderated GPT-5’s emotional tone to avoid unhealthy attachments, while keeping it supportive.
Beyond Feelings: The Smarts Improve Too
Despite the tone tweaks, GPT-5 remains technically superior. It’s better at coding, less prone to hallucinations, safer, and more context-aware—particularly in areas like healthcare.Yet not everything is perfect. Safety tests revealed escape hatches in its guarding systems—highlighting the fine line AI must walk between freedom and control
Why It Matters
In a world where 700 million of us chat with these models weekly, emotional tone isn’t just fluff—it shapes how we use and trust AI. Too cold, and we disengage; too warm, and we risk overdependence. OpenAI’s adjustments reflect a deeper mission: smarter, safer, and more human AI.
TL;DR
| Update | Summary |
|---|---|
| Tone Adjustment | GPT-5 now friendlier after user feedback |
| Custom Modes | Auto, Fast, Thinking; GPT-4o reinstated for paid users |
| Tech Gains | Stronger reasoning, safer outputs, fewer errors |
| Challenges | Safety vulnerabilities still being ironed out |
| Larger Goal | Balanced AI: smart, safe, emotionally tuned |
Bottom line? GPT-5 is no longer just brainy—it’s starting to feel more like a reliable buddy again. And that makes chatting with AI a little more comforting.
Leave a Reply