MMOOZZGG.com

Glossary

What is AI in legal work

In short

AI in legal work means programs that help lawyers search laws and case law, review and draft contracts and sort large sets of documents.

AI in industries · AI in legal

MMOOZZGG.com editorsChecked: 1 min read

For a child

Imagine the thick rule book of a big board game. When friends argue about whether a move is allowed, you spend a long time looking for the right page. A helper finds it in a second and reads it aloud. But sometimes it gets confused and makes up a rule that is not in the book. That is why adults always open the book themselves and check the page.

In plain words

A lawyer reads a lot: laws, court decisions, contracts, letters. A program based on a language model finds the right place in these texts, retells a long document, compares a contract with the company’s standard template and marks risky clauses. It also prepares the first draft of a claim or a letter. For example, before buying a company you need to read five hundred supply contracts: the helper finds those with penalties and termination rights, and the lawyer starts with them. The weak point: the model can cite a case or an article that does not exist.

For business

The gain is in faster document review, and in lawyers spending time on hard questions instead of searching. Common scenarios: checking contracts against the company’s list of requirements, due diligence in deals, answering staff about internal policies, and standard letters. The price depends on the volume of documents, the need for a closed environment and integration with the document management system. The risks: made-up citations of case law, leaks of a client’s confidential data if documents go to an outside service, and liability to the client for an error. Minimum protection: answers only from your own document base with a link to the source, and a lawyer checks every quote. Whether the cloud fits or you need your own servers is decided separately for each type of data.

For a specialist

The main classes of tasks: legal information retrieval, contract clause extraction, document classification, summarization and draft generation. Modern systems are built on large language models with RAG over a base of laws and case law. Vendors claimed that retrieval solves the hallucination problem, but the study by Magesh and co-authors (2024) showed that specialized RAG-based legal research tools hallucinate in 17–33% of answers, though less often than the general-purpose model GPT-4. Hence the practice: check quotes against the original source, limit the answer to the retrieved fragments and evaluate on a labeled set of lawyers’ questions. The European Commission classes AI in the administration of justice, for example for preparing court decisions, as high risk under the EU AI Act.

Example

The legal department of a distributor receives contracts from partners in their own wording. The helper compares each one with the standard form and shows the changed clauses on liability and payment terms, and the lawyer decides what to accept.

Related terms

Questions and answers

What AI suits lawyers? Systems that answer from a verifiable document base and show a link to the source of each quote. A general chatbot without access to legal databases is fine for drafts, but not for citing laws and case law.
Can you trust the citations a neural network gives? Not without checking. Even specialized legal tools in a 2024 study were wrong in a noticeable share of answers, so the lawyer opens every article and every case in the original source.

Sources

Where companies use it

Not sure whether your company needs this?

AI solutions map
Contact us