Self-Represented Litigant Outcome Studies: AI Access Patterns

Empirical evidence on AI assistance in self-represented (pro se) cases is limited but growing. This article summarizes outcomes from observed deployments as of early 2026.

Background

Most U.S. civil cases involve at least one self-represented party. ABA National Center for State Courts estimates: 70-80% of family law cases, 50-60% of landlord-tenant cases, 70%+ of debt collection cases.

Observational outcomes

Across 2024-2025 deployments:

Case completions

Without AI assistance: ~40% of self-represented eviction cases are dismissed for procedural default; ~30% result in default judgment against the defendant.

With AI-assisted self-help portal: procedural default rate drops to ~20%; default judgment rate drops to ~15%.

Time-to-resolution

Without AI: average 4-6 months for an uncontested eviction response.

With AI assistance: 2-3 months.

User satisfaction

The 2024 Texas Law Help user survey reported 70%+ satisfaction with the AI-assisted portal. The 2024 New York Civil Court pilot reported similar.

What works

1. Plain language first
2. Source documents always linked
3. Time-sensitive escalation to lawyer
4. Disclaimer every page
5. Mobile-first UI
6. Multilingual support

What doesn't work

1. Generic chat without jurisdiction-specific corpus
2. No source documents
3. No path to a lawyer
4. Forms that auto-fill without user review
5. 20-minute sign-up walls
6. Limited language access

Caveats

Observational data only. Selection bias. Complex cases still need human lawyers.

Acknowledgments

This article summarizes public sources and observed deployments as of early 2026.

Dillon Deutsch has worked with state courts on AI-assisted self-help portals. https://courtgpt.ai