Artificial intelligence tools have attained a notable milestone by matching or surpassing the capability of human lawyers in producing reliable contract drafts. This was documented in the initial comprehensive comparative study of AI performance against legal professionals, recently published by LegalBenchmarks.ai. According to the research, human lawyers achieved a reliability rate of 56.7% for initial drafts, whereas several AI tools achieved or surpassed this metric.
The study, titled “Benchmarking Humans & AI in Contract Drafting,” involved evaluating 13 AI tools alongside human lawyers through 30 real-world contract drafting tasks. The study’s top performer, Gemini 2.5 Pro, attained a 73.3% reliability rate, besting the highest-performing human lawyer who recorded 70% reliability. This evaluation covered output reliability, usefulness, and workflow integration, with 450 task outputs assessed and input from 72 legal professionals.
Interestingly, while specialized legal AI tools demonstrated superior performance in high-stakes legal risk scenarios, identifying risks in 83% of outputs, general-purpose tools only managed a 55% rate. Human lawyers, in these scenarios, failed to flag risks altogether. For instance, AI tools raised enforceability concerns for a potentially unenforceable penalty clause under New York law, a risk overlooked by human lawyers.
The research demonstrated significant variations in AI performance, with reliability scores ranging from 44% to 73.3% across different tools. Google’s Gemini 2.5 Pro emerged as the top performer, closely followed by OpenAI’s GPT-5.
Beyond reliability, workflow integration emerged as a critical differentiator for legal-specific platforms. A substantial portion of tested legal AI products offers seamless integration with Microsoft Word, where most contract drafting occurs. Tools such as Brackets, GC AI, and SimpleDocs led in platform workflow support, featuring libraries, clause storage, and quality assurance tools tailored for legal tasks.
The human advantage remains evident in tasks necessitating commercial judgment and context understanding, where lawyers excelled at discerning client intent and integrating multifaceted information sources. However, AI tools offered consistency in routine drafting without the temporal demands typical of human efforts, completing tasks in seconds compared to the nearly 13 minutes taken by lawyers.
In surveying 72 lawyers utilizing AI in legal work, the study revealed a preference for using multiple AI tools, with accuracy not the sole concern for adoption. Only 6% demanded complete accuracy prior to AI usage. The consensus among participants underscored verification ease and improved context management as significant for increasing AI utilization.
The findings reflect a nuanced future for legal drafting, highlighting a collaborative approach integrating the speed and accuracy of AI tools with human expertise. As noted by the researchers, “The future of drafting will not be decided by one side or one tool. It will be shaped by orchestration: combining the speed and consistency of general AI, the workflow fit of legal AI, and the judgment of lawyers.”