Why AI Won't Revolutionize Law (At Least Not Yet)

AI as a Normal Technology

• 💡 The "AI as normal technology" framework suggests that AI's societal impact, like past general-purpose technologies such as electricity, will take decades to fully materialize, not just rapid capability improvements.
• ⏳ Technology diffusion involves four stages: capability improvements, translation into products, worker adoption, and finally, organizational and legal changes, which is often the slowest and most challenging stage.
• 🧠 The legal field is a unique case study because it is purely cognitive like software engineering, yet highly professionalized and regulated like medicine, placing it in a middle ground for AI impact analysis.

Understanding Legal Service Costs

• 💰 Legal services are inherently expensive due to several structural features, including their nature as credence goods, where quality is difficult for consumers to assess even after consumption.
• ⚖️ The value of legal services is often relative and adversarial; success depends on the actions of opposing parties, not just absolute quality.
• 📜 Professional regulations, such as unauthorized practice of law (UPL) rules and entity ownership restrictions, significantly limit innovation and efficiency in the legal market.

Key Bottlenecks for AI in Law

• 🚫 Regulatory barriers, including broadly defined UPL rules and restrictions on non-lawyer ownership of legal businesses, prevent the integration of AI-driven efficiencies and new business models.
• ⚔️ Arms-race dynamics in litigation and even transactional work mean that if both sides use AI, the result is often more work and complexity, not reduced costs, mirroring the impact of e-discovery.
• 🧑‍⚖️ The need for human involvement and judgment acts as a bottleneck; judges and human lawyers have finite capacity, and there's a normative argument for human decision-makers in critical legal processes.

The Role of Human Judgment

• 🎯 The paper concedes that AI can perform many complex cognitive legal tasks, but the skepticism about cost reduction stems from systemic issues, not AI's raw capabilities.
• 🌍 A core argument is the normative commitment to human agency in high-stakes decisions, viewing it as fundamental to controlling our civilization and making societal choices, rather than delegating to machines.
• ✅ While human judgment is crucial for high-stakes cases, AI can still provide significant benefits for access to justice and automating administrative tasks that fall outside strict regulatory barriers.

Proposed Reforms and Benefits

• 🛠️ Reforms like regulatory sandboxes (e.g., in Utah) and regulatory markets are proposed to allow for innovation and new models of legal service delivery without immediately dismantling existing protections.
• 🌱 By addressing these bottlenecks, AI can help modernize the legal system, making it more efficient and expanding access to legal services for those who currently cannot afford them.
• 🚀 The goal is not to keep AI out, but to strategically integrate it by understanding and reforming the institutional structures that currently impede its positive, long-term transformative potential.