A contract can preserve legal precision while making obligations easier to find and understand. The challenge is to remove avoidable difficulty without deleting qualifications, changing risk allocation or hiding uncertainty. Generative AI can propose alternatives, but clarity and legal equivalence still need testing.
Treat the magic spell as a metaphor
Helena Haapio's 2025 chapter “No More Legal Writing in Contracts” uses the idea of a magic spell to challenge inherited drafting conventions. It is a design argument and a useful prompt to ask which inherited expressions still serve a legal function; it is not evidence that technical terms are useless.
A term may carry a specific legal function. Identify that function before changing it. Do not assume that plain wording, a diagram or an AI rewrite automatically satisfies the applicable formalities or preserves an established interpretation.
What comprehension studies show
In Martínez, Mollica and Gibson's 2022 research, difficult linguistic features reduced recall and comprehension of contract excerpts. Clauses embedded in the middle of another clause were particularly problematic for recall. The findings support reducing avoidable syntactic complexity, but do not establish that every shorter contract is clearer or legally equivalent.
Their 2023 experiments with US lawyers also found better comprehension of simplified text and favourable ratings of simplified contracts. Participants' ratings of enforceability were perceptions, not judicial rulings. This distinction matters when using research to justify a drafting change.
Check meaning before approving a rewrite
Start with a clause whose purpose the team understands. Record the actor, required action, trigger, deadline, exceptions and consequences. Compare each element after the rewrite. Check defined terms, cross-references, amendments and any rule that requires particular form or wording.
For example, splitting a long conditional sentence may improve navigation but accidentally make an exception apply to the wrong obligation. Treat that as a test case. A prompt such as “simplify without changing meaning” expresses an instruction; it does not verify that the instruction was followed.
Use layout and visuals for a specific task
Headings, timelines and role-based tables can help readers locate information. The 2017 contract-diagram study by Passera and colleagues examined comprehension among contract professionals, including native and non-native English speakers. Its reported benefits concern the tested documents and tasks, not every visual contract or an AI-generated design.
Decide whether a summary is explanatory or part of the agreement, and handle inconsistencies explicitly. Keep the relevant conditions available where a user acts. A simplified front page is unhelpful if it hides the exception that determines the answer.
Give AI a bounded drafting role
Use only material approved for the environment. Ask for alternatives, a list of changed terms and clause references. Require the model to flag ambiguity rather than silently resolve it. Check every important change against the source and applicable law.
Test the result with the intended readers and languages. Ask them to find an obligation, explain a condition and identify the next action. Measure correct answers and misunderstandings, not just whether they prefer the appearance. An AI-generated persona cannot replace feedback from the people who perform the contract.
Plain-language contract review
0/6Key Takeaway