Los Angeles is testing whether AI can help clear clogged court dockets — a development crypto projects and litigants should watch as legal systems strain under rising caseloads and more automated filing tools.
What’s happening
- The Los Angeles Superior Court has launched a pilot giving a small group of judicial officers access to an AI system called Learned Hand. The tool summarizes filings, organizes evidence, and generates draft rulings in civil cases, handling administrative “drudge work” so judges can focus on legal analysis and discretion.
- Learned Hand was founded in 2024 by Shlomo Klapper, a former U.S. Court of Appeals law clerk and Palantir deployment strategist. The company is named after the noted federal judge Learned Hand.
Why the pilot now
- Courts face mounting pressure: national law firm Fisher Phillips reported filings rose 49% in the past year — from an average of 4,100 to 6,400 — partly because AI makes it easier and cheaper to produce filings.
- “We’re at a place in society where courts are under tremendous strain,” Klapper told Decrypt. He said advances in AI are “massively dropping the cost of litigation,” increasing the number of filings courts must handle.
How Learned Hand works and safeguards
- The pilot tests the system across a case lifecycle, from intake to draft rulings. Klapper says the aim is to surface key facts and legal issues while leaving final judgment entirely to human judges.
- Presiding Judge Sergio C. Tapia II emphasized the limits: the tool “may enhance the way judicial officers review and engage with case files and information, it will not replace, or in any way compromise, the sanctity, independence, and impartiality of judicial decision-making.”
- Learned Hand’s design decisions target a common AI failure mode: hallucinations. Klapper noted that the real cost is not generating text but verifying that output against case materials and legal sources: “Most of the expense of our large language model is in the verification, not the generation.”
- To reduce risk, Learned Hand operates on a narrower, defined set of legal materials rather than scraping the open internet. It breaks tasks into steps and assigns each step to a model with a specific function. The interface is built to be nontechnical — “point and click” — so judges don’t need prompt engineering expertise.
- Klapper is explicit about limits: judges “should not take AI outputs at face value” and “don’t trust, verify.”
Why this matters (and the risks)
- AI errors in legal settings have already made headlines: in 2023, the defense in the Pras Michel case alleged an AI-assisted closing argument contained frivolous claims, and a federal judge ordered lawyers for Michael Cohen to provide printed copies of cited cases after the court could not verify them.
- For industries like crypto — where disputes often involve technical evidence, rapid innovation, and a surge in filings — more efficient case processing could be helpful. But accuracy, transparency, and human oversight will be critical to avoid amplifying mistakes or bias from underlying training data.
Bottom line
The LA pilot is a cautious, human-first experiment in offloading routine judicial tasks to AI while keeping final decisions with judges. If it proves reliable, it may offer a blueprint for courts handling growing volumes of litigation — including the complex, fast-moving disputes that frequently appear in the crypto space.
Read more AI-generated news on: undefined/news
