Certifying Legal AI Assistants for Unrepresented Litigants: A Global Survey of Access to Civil Justice, Unauthorized Practice of Law, and AI
Notice bibliographique
Résumé
The global integration of artificial intelligence (AI) into legal services has created a critical need for clarity regarding unauthorized practice of law (UPL) rules. Traditionally, UPL rules prohibited unlicensed individuals from engaging in activities legally reserved for qualified attorneys, including, in some jurisdictions, offering legal advice, interpreting laws, representing clients in court, or drafting legal documents. Now that some AI systems can perform functions that practice of law regulating authorities have traditionally reserved for licensed attorneys, a framework is needed to certify the use of legal AI assistants by unrepresented litigants. Ensuring the accuracy of information provided by legal AI assistants for unrepresented litigants benefits the entire legal community, including attorneys, by promoting stricter standards and higher acceptance thresholds. We examine the perspectives of several primary stakeholders in certifying legal AI assistants, including unrepresented litigants, practice of law regulating authorities, judiciaries, the legislature, the legal aid community, and the legal tech community. We conduct a detailed survey of access to justice, AI, and UPL in various international jurisdictions, including Argentina, Australia, Brazil, Canada, China, the European Union, Germany, India, New Zealand, Nigeria, Singapore, the United Kingdom, and the United States. In each of these jurisdictions, we explore how UPL is currently managed in the context of legal AI use by unrepresented litigants. We also include a 50-state and 6-territory survey for the United States on what each Bar Association and Judiciary is doing to regulate legal AI use by unrepresented litigants. In light of this survey, we propose that practice of law regulating authorities add certified legal AI assistants to their lists of UPL exemptions so that such assistants can provide specific and useful legal information, guidance, and advice to unrepresented litigants. We propose a capability-based framework for certifying legal AI assistants for unrepresented litigants. This is intended as a harmonized global proposal, designed for local implementation by each jurisdiction’s practice of law regulating authority, with the flexibility to address individual jurisdictional nuances. Unrepresented litigants are already using AI chatbots for help in legal proceedings, sometimes to their detriment. Our proposal aims to allow unrepresented litigants to use legal AI assistants that have been verified for accuracy. This framework addresses the key justification for UPL restrictions—the risk of incorrect legal guidance—by basing the certification of individual capabilities on their accuracy when tested on public benchmark datasets. Legal AI assistants are added to lists of UPL exemptions under this approach if their accuracy meets or exceeds a certification threshold when tested on these public benchmark datasets. The jurisdiction’s practice of law regulating authority would set the certification threshold or, as we suggest, a third-party certifying authority delegated to perform this task. While many public benchmark datasets are required under this framework, the legal AI community is rapidly developing such datasets. To enable AI to enhance access to justice for unrepresented litigants globally, practice of law regulating authorities in each jurisdiction must choose to exempt certified legal AI systems for unrepresented litigants from unauthorized practice of law regulations.
Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.
Comment cette classification a été obtenuedéplier
Prédiction machine sur la base complète
Imitation des enseignantsNi prévalence calibrée, ni vérité terrain. Validation humaine à venir. Le volet Gemma est une étiquette directe du modèle pour chaque travail de la base, lue sur la notice réduite au titre. Le volet Codex est un classifieur appris des 10 348 étiquettes directes de Codex et calibré sur les taux pondérés de l'échantillon; les champs sans appui suffisant ne portent aucun appel Codex. Le mode candidate est l'union des deux volets; le consensus est leur intersection. Ces sorties portent le statut machine_predicted_unvalidated et ne sont pas des étiquettes humaines.
Scores du classifieur distillé par catégorie (deux têtes)
| Catégorie | Codex | Gemma |
|---|---|---|
| Métarecherche | 0,003 | 0,010 |
| Méta-épidémiologie (sens strict) | 0,000 | 0,000 |
| Méta-épidémiologie (sens large) | 0,000 | 0,000 |
| Bibliométrie | 0,003 | 0,003 |
| Études des sciences et des technologies | 0,001 | 0,002 |
| Communication savante | 0,002 | 0,004 |
| Science ouverte | 0,001 | 0,003 |
| Intégrité de la recherche | 0,001 | 0,001 |
| Charge utile insuffisante (le modèle a refusé de juger) | 0,003 | 0,000 |
Scores machine (provisoires)
Les deux têtes enseignantes du modèle étudiant, lues sur ce travail. Un score ordonne la base pour la relecture; il n'affirme jamais une catégorie, et le statut de validation accompagne chaque rangée tel quel.
Scores de référence d'un modèle non mature (critères de maturité non atteints, 7 itérations). Un score ordonne; il n'affirme jamais une catégorie.
score_only:v0-immature-baseline · tel quel depuis la passe de notation : score_only signifie que le nombre peut ordonner les travaux, et qu'aucune étiquette de catégorie n'en découleClassification
machine, non validéePrédiction automatique; un appel candidat d’une seule source (Gemma direct ou Codex distillé), pas un consensus.
Le détail, modèle par modèle et score par score, se trouve en fin de page sous « Comment cette classification a été obtenue ».