This website uses cookies
Read our Privacy policy and Terms of use for more information.
Benchmarks
+2
Oct 15, 2026
•
4 min read
Why a trustworthy AI evaluation needs a way to say: no valid result.
Judgment
Sep 22, 2026
1 min read
A short explainer video on how LLMs produce an answer, and why fluent is not the same as true.
+1
Sep 16, 2026
Human-first drafting can preserve independent engagement before AI enters the workflow. It cannot, by itself, establish where the final judgment came from.
Sep 9, 2026
3 min read
Completeness is a virtue in prose and a position in a legal instrument. The tool cannot tell the two apart.
Aug 27, 2026
An assessment can become stale without ever having been wrong.
Aug 21, 2026
A court just identified the half of AI review that human verification does not reach.
Aug 10, 2026
Every edit was justified. The paragraph was built around the wrong point.
+3
Aug 3, 2026
How a benchmark can stay accurate while becoming misleading.
Jul 30, 2026
Human review matters. It is also being asked to prove things it cannot establish.
Jun 10, 2026
2 min read
Competence is shifting from "is the output correct" to "can you show where your judgment took control." A short record closes most of the gap.
Jun 8, 2026
The AI mistakes that get pro se litigants mocked are happening inside expert legal work too. They are just better dressed, which is what makes them dangerous.
Jun 7, 2026
Why legal competence now requires lawyers to preserve friction, not just verify outputs.
