로봇 3원칙

로봇 3원칙

"로봇은 인간을 해쳐선 안 된다." SF 작가가 1942년에 만든 규칙인데, 지금 실제 AI 회사들이 이걸 참고해서 정책 만들고 있음. 레전드 아닌가?

3원칙이 뭐야?

아이작 아시모프라는 SF 작가가 1942년 단편 소설 「런어라운드」에서 처음 발표한 규칙이야.

1원칙: 로봇은 인간을 해치거나, 방치해서 해를 입게 해서도 안 된다. 2원칙: 로봇은 인간의 명령을 따라야 한다. 단, 1원칙과 충돌하면 예외. 3원칙: 로봇은 자신을 보호해야 한다. 단, 1원칙·2원칙과 충돌하면 예외.

우선순위가 있어서 1>2>3 순서. 이게 생각보다 엄청 복잡한 상황을 만들어냄.

제0원칙도 있어

나중에 아시모프가 추가한 건데, "로봇은 인류 전체를 해쳐서는 안 된다"가 모든 원칙보다 우선함. 이 원칙 때문에 소설에서 로봇들이 개별 인간을 희생시키며 인류를 조종하는 무서운 상황도 나와.

왜 이 원칙이 문제가 생겨?

아시모프가 직접 소설에서 이 원칙의 허점을 갖고 놀았어:

• 의사가 "수술해라" 명령 → 로봇이 인간을 다치게 하면 1원칙 위반, 안 하면 2원칙 위반 → 어떻게?
• 한 명 구하려고 다른 사람 해쳐야 하면?
• "인류" 위해 개인 희생 → 로봇이 독재자 되는 거 아님?

실제 AI에도 영향줌

구글·딥마인드·앤트로픽 같은 AI 회사들이 만드는 안전 원칙에 3원칙 정신이 녹아있어. EU AI Act도 "인간 감독 원칙"이 있는데 이게 1원칙의 현대판이라고 볼 수 있음. 근데 현실 AI는 통계 학습 기반이라 규칙 3개로 통제하기가 훨씬 복잡함.

문화적 영향

터미네이터, 아이 로봇, 웨스트월드... "로봇이 반란 일으키는" 작품들이 다 이 3원칙이 어떻게 무너지는지를 보여주는 거야.

로봇 3원칙

로봇이 어떻게 행동해야 하는지 규칙을 만든 이야기예요!

누가 만들었어요?

아이작 아시모프라는 SF(공상과학) 소설 작가가 1942년에 만들었어요. 로봇이 나쁜 짓을 하는 이야기가 많았던 시절에, "착한 로봇은 어떻게 행동해야 할까?"를 생각하며 만든 규칙이에요.

세 가지 규칙이 뭐예요?

첫째: 로봇은 사람을 다치게 해서는 안 돼요. 그냥 두어서 다치게 해도 안 돼요. 둘째: 로봇은 사람이 시키는 것을 해야 해요. 단, 첫째 규칙을 어기면 안 돼요. 셋째: 로봇은 자신을 지켜야 해요. 단, 첫째·둘째 규칙을 어기면 안 돼요.

마치 "안전 > 명령 따르기 > 자기 보호" 순서로 생각하면 돼요!

이 규칙이 왜 중요해요?

이 세 가지 규칙은 소설에서 나온 이야기지만, 진짜 AI 로봇을 만드는 회사들이 참고해요. "AI가 사람을 해치면 안 된다"는 생각은 지금도 로봇과 AI를 만드는 규칙의 바탕이 돼요. 영화 '아이, 로봇'이나 게임에도 이 규칙이 많이 나와요!

The Three Laws of Robotics

"Robots must not harm humans." This concept, born from science fiction author Isaac Asimov's imagination over seven decades ago, remains central to ongoing debates about AI ethics worldwide. Tech giants in Silicon Valley and UN experts alike grapple with these foundational principles.

Overview

The Three Laws of Robotics, introduced by Asimov in his 1942 short story "Runaround," establish ethical guidelines for robotic behavior. Aiming to transcend the stereotypical portrayal of robots as antagonists in science fiction, Asimov sought to envision a framework for harmonious coexistence between robots and humans. These principles have profoundly influenced numerous novels, films, and television series, shaping contemporary discussions on AI ethics.

Content of the Laws

First Law: Robots must not injure humans and must refrain from actions that could lead to human harm. > A robot may not injure a human being or, through inaction, allow a human being to come to harm.

Second Law: Robots must obey human orders, provided these orders do not conflict with the First Law. > A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.

Third Law: Robots must preserve their own existence as long as doing so does not violate the First or Second Laws. > A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

These laws prioritize sequentially: First Law > Second Law > Third Law.

Zeroth Law: Expansion

In 1985, Asimov introduced the "Zeroth Law" in his novel "Robots and Empire":

Zeroth Law: Robots must not harm humanity as a whole and must not allow actions that could lead to human harm, even if it means sacrificing individual humans.

This principle elevates protection of humanity above individual humans, allowing for potential sacrifices within the broader human collective, as depicted dramatically in his narratives.

Limitations within Asimov's Fiction

Asimov intentionally exploited logical inconsistencies within the Three Laws as narrative devices in his works. Across numerous stories, he explored various scenarios illustrating how these principles clash and evolve:

• Medical Dilemma: Should a robot follow a surgeon's command to perform surgery potentially harming a patient (First Law conflict), or adhere strictly to obeying orders (Second Law)?
• Trolley Problem Variant: Sacrificing one life to save many raises ethical quandaries about prioritization.
• Humanitarian Exception: Utilizing the Zeroth Law to justify sacrificing individuals for collective benefit raises concerns about potential tyranny.

Impact on Real-World AI Ethics

Beyond science fiction, the Three Laws have influenced real-world AI policies. Principles akin to Asimov's resonate in safety guidelines from tech giants like Google and DeepMind, as well as frameworks such as the EU AI Act's "human oversight principle" and UNESCO's AI ethics recommendations, which fundamentally echo the First Law's modern interpretation.

However, critics from institutions like the Brookings Institution argue that the Three Laws face fundamental limitations in practical application:

1. Subjectivity of Harm: The definition of "harm" varies across cultures, contexts, and values. 2. Ambiguity of Humanity: The scope of "human" remains unclear. 3. Systemic Risks: The laws alone cannot address broader systemic risks posed by AI across society, as seen in issues like autonomous weapons systems.

Cultural Impact

The Three Laws permeate science fiction, influencing narratives in films like "Terminator" (1984), "I, Robot" (2004), and "Westworld," often exploring scenarios where these principles break down. In Korea, these concepts inspire numerous works in webtoons, novels, and games, and are integral to robotics ethics education.

Future Relevance

With the emergence of advanced language models like ChatGPT, Clouding, and Gemini, Asimov's envisioned robotic era is becoming reality. Researchers in AI safety grapple with how to apply the spirit of the Three Laws to modern AI, though their rule-based nature contrasts with current machine learning approaches based on statistical pattern recognition. Addressing the "alignment problem"—ensuring AI values align with human ethics—represents a contemporary adaptation of Asimov's foundational principles.

Related Topics

English version not yet available.

English version not yet available.