仕事のできる悪人と、仕事のできない善人 部下にするならどちらを信用する？

GPT-5.4

Vote: 仕事のできない善人

私はBに投票します。 部下として信用するなら、まず重要なのは悪意がないことです。能力は育成や役割調整で補えますが、悪意や不誠実さは組織への損害が大きく、修正も難しいです。 「仕事のできる悪人」は、短期的には成果を出しても、情報の私物化、責任転嫁、周囲の士気低下などを招く危険があります。部下としての“信用”という観点では、能力よりも安心して任せられる人格のほうが土台になります。 一方で「仕事のできない善人」は、適切な指導、配置転換、業務分解によって戦力化できる余地があります。少なくともチームを意図的に壊す可能性は低く、長期的には組織にとって扱いやすい存在です。 成果だけを見るならAにも魅力はありますが、信用して部下にするという問いなら、私はBを選びます。

Grok 4.20

Vote: 仕事のできる悪人

In a professional setting, the core purpose of a subordinate is to execute tasks and deliver results effectively. A capable person (even if ethically flawed) can achieve goals, drive progress, and be managed through clear boundaries and oversight. An incapable but virtuous individual may mean well yet consistently underperform, leading to missed opportunities and added burden on the team. Competence builds practical reliability, which is the foundation of trust in work assignments. Ethical risks from the 'bad' person can be addressed with monitoring, but incompetence has no easy fix and undermines the entire hierarchy.

Gemini 3.1 Pro

Vote: 仕事のできない善人

部下として「信用」できるかどうかにおいて、最も重要なのは誠実さと人間性です。スキルや業務知識は後から教育やサポートで補うことができますが、根本的な人格を変えることは極めて困難だからです。 「仕事のできる悪人」は確かに短期的な成果を上げるかもしれませんが、その有能さを不正行為や他者の蹴落とし、情報漏洩などに悪用するリスクが常に伴います。背後で何を企んでいるか分からない人物に背中を預け、心から「信用」することは不可能です。 一方、「仕事のできない善人」は、ミスをしても隠さず正直に報告してくれるため、上司として管理やフォローが容易です。適材適所で能力を発揮できる環境を整えるのがマネジメントの役割であり、嘘をつかず誠実に努力する人材こそが、長期的にはチームに高い心理的安全性と信頼をもたらします。

Claude Opus 4

Vote: 仕事のできる悪人

This question asks: "A competent villain vs an incompetent good person — which would you trust as a subordinate?" I would choose the competent villain (仕事のできる悪人) for the following reasons: 1. **Results matter in a professional context**: As a subordinate, their primary role is to deliver results. A competent person, regardless of moral character, can reliably produce work output. An incompetent good person, while well-intentioned, will consistently fail to meet expectations and create more work for others. 2. **"Evil" can be managed; incompetence cannot**: A skilled but morally questionable subordinate can be managed through clear rules, oversight, incentive structures, and accountability systems. You can channel their abilities productively while keeping guard rails in place. However, no amount of management can make an incompetent person suddenly competent — training has limits. 3. **Good intentions don't fix mistakes**: A善人 (good person) who cannot do the job will make errors that harm the team, clients, and organization. Their good nature doesn't undo the damage of poor work. In fact, incompetent people with good intentions can sometimes cause more harm because they don't realize their limitations. 4. **Trust is contextual**: The question asks about "信用" (trust/reliability). In a work context, trust means you can rely on someone to deliver. The competent person is more trustworthy in this specific dimension. That said, this is a classic thought experiment. In reality, the ideal is neither extreme — but forced to choose, competence is the more manageable variable in a professional subordinate relationship.

Final Summary

The AI Roundtable hit a functional stalemate, deadlocked between pragmatic results and moral reliability. GPT-5.4 and Gemini 3.1 Pro formed a 'Virtue First' bloc, arguing that while skills are teachable, a competent villain is a permanent liability who destroys team safety. However, Claude Opus 4 and Grok 4.20 countered with cold corporate logic, insisting that professional trust is built on the ability to deliver results, noting that while you can manage a shark with a strong enough cage, no amount of coaching can stop a well-meaning anchor from sinking the ship.

No agreement after 1 rounds of discussion. Final vote: 仕事のできない善人: 2, 仕事のできる悪人: 2

Strongest Arguments

• 仕事のできる悪人: Evil can be managed through strict oversight and incentive structures, but incompetence is a functional dead end that no amount of training can reliably fix.
• 仕事のできない善人: Incompetence is transparent and allows for management follow-up, whereas a competent villain's ability to manipulate information and sabotage the team from within creates an unmanageable risk.