Human-In-The-Loop Legal AI

Human-in-the-loop legal AI is the use of legal AI where a human lawyer, court actor, or responsible professional remains accountable for judgment, verification, and advocacy. In Continental Rift: NATO’s Tense Summit, Anna Kerr presents this as the constructive alternative to Vibe Lawyering and Legal AI Hallucination.

The source’s example is Garfield AI, which helped a user recover money in a London court while a human lawyer argued the case. The broader point is that AI may improve access to justice by lowering drafting and preparation costs, but courtrooms remain conversational, adversarial, and responsibility-bearing environments where generic chatbot output cannot be treated as legal judgment.

Key Claims

• Legal AI can expand access only if it improves preparation without hiding responsibility.
• Human review must include citation checking, claim selection, settlement judgment, and courtroom advocacy.
• Domain-specific legal systems are more plausible than general chatbots for high-stakes legal work.
• Courts may need rules that distinguish assistive drafting from unverified AI substitution.

Connections

• Garfield AI - source example.
• Anna Kerr - source participant explaining the distinction.
• Vibe Lawyering and Legal AI Hallucination - risks the concept responds to.
• AI Governance And Compliance, Human Judgment Under AI, and Output Quality Gates - adjacent control and review frames.