AI vs. human capability
AI vs. Human Education Level
The years of education AI output is judged equivalent to, task by task, versus humans.
Across ten legal tasks, HAQQ's benchmark rates AI output at 14.8-15.7 years of education-equivalent, with human performance close behind at 12.5-15.3 years. Most gaps are narrow: on examining records for compliance, AI and humans tie at 15.3 years. The outlier is explaining tax laws, where AI rates 14.9 years against a 12.5-year human baseline, a 2.4-year gap wider than any other task. For legal teams, this suggests AI matches human-level output on core research and drafting, with the largest edge in specialized regulatory explanation rather than general legal reasoning.
Os dados
| Task | AI (years) | Human (years) |
|---|---|---|
| Read & rewrite legal documents | 14.8 | 14.2 |
| Determine liability per law | 15.5 | 15.1 |
| Research case law | 15.7 | 15.3 |
| Examine legal documents for court | 15.2 | 14.7 |
| Examine records for compliance | 15.3 | 15.3 |
| Explain tax laws | 14.9 | 12.5 |
| Research laws for hearings | 15.1 | 14.8 |
| Review legal publications | 15 | 14.5 |
| Research legal issues & write opinions | 15.4 | 15 |
| Interpret laws, rulings, regulations | 15.2 | 14.9 |
Dados medidos - pré-processados a partir do conjunto de dados de uso real do Legal AI Index da HAQQ (mais de 134.000 pontos de dados em 30 países) ou do benchmark jurídico cross-model próprio da HAQQ /50.
Pesquisa relacionada
- Tasks That Still Need a HumanThe share of each legal task that professionals say still requires human judgment.
- AI Autonomy Scores by TaskHow autonomously AI can handle each legal task, on a 1-5 scale with confidence intervals.
- Legal Productivity BenchmarksBefore-and-after time per workflow, with the productivity gain from AI.