Trust in Legal AI Grows with Experience, American Arbitration Association® and Jus Mundi Study Finds

A survey of 557 U.S. arbitration professionals found daily AI users reported trust scores of 3.42 out of 5, versus 0.83 among non-users, indicating practical experience materially improves confidence in legal AI tools. Respondents expect AI to absorb routine tasks such as document review (51%) and cite-checking (47%), while 52% expect strategic legal judgment to become more valuable. Accuracy and hallucinations remain the leading concern for 78% of respondents, and practitioners prioritize institutional AI guidelines (30%) and arbitration-specific tools (24%) to support responsible adoption.

Analysis

This is a weak near-term public-equity signal, but it reinforces a more investable legal-software bifurcation: workflow incumbents with privileged legal content, embedded distribution, and enterprise-grade data controls should monetize before general-purpose model vendors. RELX (RELX), Thomson Reuters (TRI), and Wolters Kluwer (WKL.AS) can package AI into existing research, drafting, and compliance workflows with low customer-acquisition cost; the relevant upside is retention and ARPU expansion rather than immediate seat-count growth. Smaller legal-tech vendors lacking proprietary content or defensible distribution face feature commoditization as document review, citation checking, and research become baseline capabilities.

The highest-value bottleneck is likely governance, not model quality. Persistent concern around output reliability and sensitive case materials favors vendors that provide source-linked answers, audit trails, permissioning, private deployments, and indemnification; this should support premium pricing and reduce churn for established platforms over the next 6-18 months. It also creates a second-order opportunity for cybersecurity and information-governance vendors—Palo Alto Networks (PANW), CrowdStrike (CRWD), and Microsoft (MSFT)—if law firms and corporate legal departments move sensitive workloads from ad hoc public-model use to controlled enterprise environments.

Consensus may overstate near-term labor displacement. Legal services are a trust- and liability-constrained market, so initial AI gains are more likely to be absorbed through fixed-fee capacity expansion, faster case throughput, and competitive fee pressure than outright headcount reduction. The key 1-3 month catalyst is not another adoption survey but AI-specific product KPIs and disclosed attach rates at RELX/TRI earnings; absent evidence of paid usage, the narrative alone should not justify multiple expansion.

Key Decisions for Investors

• Maintain a 6-18 month overweight in RELX and TRI versus broad software: favor incumbents with proprietary legal corpora and workflow distribution; target a 10-15% relative return if paid AI attach rates lift revenue retention or organic growth by 100-200 bps. Falsifier: two consecutive quarters without AI-linked ARPU, retention, or guidance improvement.
• Use WKL.AS as a lower-beta European expression of governed professional AI adoption; accumulate on market-driven weakness rather than chase survey-related sentiment. Thesis requires evidence that AI product investment is preserving margins while supporting subscription price realization.
• Watch-list, not recommendation: long PANW or MSFT versus a basket of smaller application-software names if enterprise legal AI deployments begin requiring private-model, identity, and data-loss-prevention spend. Confirm through legal-sector bookings commentary; without disclosed vertical demand, this is too diffuse for a standalone trade.
• Avoid positioning for broad legal-industry labor disruption over the next 12 months. Any short thesis in staffing or legal-services exposure requires evidence of declining billable hours, weaker pricing, or accelerated associate hiring cuts—not user-intention data.