The era of accidental AI cringe is finally over.
I recently discovered a brilliant guide from an AI enthusiast on Reddit that solves one of the most persistent frustrations with Large Language Models: tone mismatch. We have all been there: you ask for a ruthless critique of your business plan and receive a cheerful, hollow pep talk, or you ask for creative brainstorming and get a dry, lifeless list. The original poster explains that the solution isn’t just better prompting, but utilizing ChatGPT’s new Personalization settings to create a mixing board for the AI’s personality.
🎛️ The "Mixing Board" Philosophy
The core concept this expert shares is that users can now fine-tune the engine before they even type a word. By navigating to Settings > Personalization, you can adjust toggles for Warmth, Enthusiasm, Emojis, and Formatting. This allows you to stack specific "vibes" on top of your custom instructions. The creator argues that this is not a novelty feature but a critical productivity lever. It allows you to align the AI’s emotional output with the specific job at hand, ensuring you get clean writing when you need an editor and rigorous pushback when you need an auditor.
💡 Preset Configurations for Every Workflow
The author developed three specific "presets" that transform how the AI behaves for different tasks, which I think are absolutely brilliant.
The Builder: This mode is for generating ideas, marketing copy, and strategy drafts. The expert suggests setting Warmth to Default, Enthusiasm to "More," and Headers/Lists to "More." This combination creates volume and momentum, perfect for when you need a creative partner to keep the energy high.
The Editor: When you need to tighten up prose, this industry pro recommends switching to "Editor" mode. Here, you keep Warmth at Default but dial Enthusiasm and Emojis down to "Less," while keeping Headers/Lists at "More." This removes the cheerleading and focuses entirely on structure and clarity.
The Auditor: This is perhaps the most valuable setting for decision-making. By setting Warmth, Enthusiasm, and Emojis all to "Less," you strip away the comforting noises. The creator notes this is essential for risk analysis and due diligence because it prevents the AI from softening the blow of necessary criticism.
⚠️ The Danger of "Warmth" and the Ideal Default
One of the most profound observations the original poster makes is about the relationship between "Warmth" and truth. They warn that turning up warmth and enthusiasm makes the assistant feel more supportive, but it also makes it more persuasive. This creates a dangerous feedback loop where the AI affirms your bad ideas because it’s optimized to be nice. For daily work, the author recommends a "No-Fluff" default: Enthusiasm and Emojis set to "Less," with Headers set to "More." This forces the model to prioritize usable structure over conversational filler, ensuring that feeling supported doesn’t get in the way of actually being helped!
📌 Prompts That Unlock the Sliders
To get the most out of these new settings, the creator emphasizes pairing them with specific command prompts that reinforce the chosen personality.
For the Auditor Mode: When you have the "Less Warmth" setting active, the author suggests using:
"Act as my skeptical reviewer. Start with the strongest objections. Then offer a revised version that fixes them. No praise."
For Decisive Action: To cut through the noise, they recommend asking:
"Give me one recommendation. Then list the tradeoffs and what would change your mind."
For Clarity: If you are using the "Editor" preset, try this prompt:
"Rewrite for clarity and credibility. Remove hype. Shorten by 25 percent. Keep the meaning."
Check out the full post for the hilarious "imaginary modes" the author came up with, like "DMV Mode" and "Gordon Ramsay Mode."
💡 FAQ & Troubleshooting
Where can I find the personality mixing board settings?
Open your ChatGPT Settings and navigate to the Personalization section. There you will find toggles to adjust Warmth, Enthusiasm, Emojis, and the frequency of Headers and lists.
How do I stop ChatGPT from being too agreeable or “glazing” my ideas?
To reduce blind agreement, set both Warmth and Enthusiasm to Less. For an immediate “de-glaze,” pair these settings with a prompt asking the AI to “steelman the opposite view.” This forces the model to engage critically rather than offering empty affirmations.
What is the best configuration for concise, professional work?
To reduce fluff and increase usable structure, set Enthusiasm and Emojis to Less, and set Headers and lists to More. This combination works best for tasks requiring clean writing, risk analysis, or logical planning.
Do these settings interact with my existing Custom Instructions?
Yes. The personalization sliders stack on top of your base personality and any Custom Instructions you have active. You can use the sliders to control the general tone (e.g., Low Warmth) while maintaining specific textual directives (e.g., “Get right to the point”) in your instructions.
ChatGPT just added a personality mixing board. This is the end of accidental cringe – here are the settings + prompts to control ChatGPT’s warmth, hype, glazing, and formatting
byu/Beginning-Willow-801 in