New AI Skill Forces Machine to Debate Decisions from All Sides

AI Advice Gets a Rigorous Overhaul

In a breakthrough for structured AI reasoning, a developer has created a skill that forces generative models to systematically examine decisions from multiple perspectives before offering recommendations. The tool, built by Juan Allo, applies Edward de Bono's Six Hats thinking framework to prevent the vague, biased advice that plagues most AI conversations.

"I noticed that AI often gives advice that confirms what you already think," said Allo, the creator of the skill. "I wanted to force it to see all angles." The result is a three-round debate that produces a concrete recommendation, key agreements, unresolved tensions, and next steps—ending the era of generic 'it depends' responses.

The Problem with Freeform AI Advice

Most AI interactions follow a simple pattern: ask a question, receive an answer, maybe ask a follow-up. While helpful, this freeform approach lacks rigor. The model tends to generate options without systematically tearing them apart, leading to confirmation bias or shallow pros-and-cons lists that don't actually clarify decisions.

Allo identifies the core issue: AI excels at generating possibilities but struggles to evaluate them critically. Users get "it depends" answers, emotional confirmation of their own viewpoints, and lists that fail to drive action.

Background: The Six Hats Framework

Edward de Bono's Six Hats method is a structured thinking technique that forces consideration of six distinct perspectives. Each hat represents a different focus:

• White Hat — Facts, data, and objective information
• Red Hat — Emotions, intuition, and gut feelings
• Yellow Hat — Benefits, upsides, and positive aspects
• Black Hat — Risks, objections, and downsides
• Green Hat — Alternatives, creativity, and new ideas
• Blue Hat — Synthesis, moderation, and overall process control

The framework has been used for decades in corporate strategy and group decision-making. Allo's adaptation applies it to AI, ensuring the machine considers every angle before concluding.

How the Skill Works

Users simply instruct the agent: "Run a six hats debate on [your decision]." The AI then runs three rounds of structured dialogue, simulating each hat in turn. After the debate, it outputs a markdown file containing a final recommendation (Blue Hat), key agreements among the perspectives, unresolved tensions that remain, and actionable next steps.

"It's not just a list of pros and cons," Allo explained. "It's a genuine debate that surfaces trade-offs and forces a real conclusion."

What This Means for Decision-Makers

This skill transforms AI from a passive answer provider into an active reasoning partner. For professionals facing complex career moves, business strategies, or product decisions, it offers a disciplined method to surface blind spots and avoid hasty choices.

"Most AI advice is generic because the machine doesn't challenge itself," said Allo. "By imposing a structure like Six Hats, we get recommendations that are specific, balanced, and actionable." The system's ability to document unresolved tensions also helps users identify areas requiring further research or human judgment.

Real-World Test: Career Pivot

In a test run, the skill tackled the question: "Should I stay in frontend tooling or move to an AI company?" After three rounds, the AI produced a nuanced recommendation: "Phased optionality — not an immediate pivot. Ship a small AI UI experiment, run exploratory interviews, talk to your current employer about AI-adjacent growth, then judge any offer on team and day-to-day work rather than the 'AI company' label."

This is a concrete, strategic suggestion—far removed from the typical generic advice AI provides. The output also included key agreements (e.g., both hats agreed AI skills are valuable) and an unresolved tension around timing versus urgency.

Availability and Next Steps

The skill is open-source and available for immediate use. Developers can clone it from github.com/juanallo/six-hats-skill, add it to their agent skills folder, and run debates on any decision. Allo encourages others to contribute improvements and expand the framework to other structured thinking methods.

"This is just the beginning," he said. "We can apply the same approach to SWOT analysis, decision trees, or even negotiation planning. The goal is to make AI advice truly useful—not just fast."