How can AI use legal knowledge in ways that meet the demands of legal practice? This question is central to Legal Cognition. What matters is how models, legal sources and an organisation’s knowledge work together: How do agents receive the relevant context? How can professional requirements guide their work, and how can people trace and review their results? Our goal is infrastructure in which legal expertise determines how AI works – while people retain professional responsibility.
We are developing a benchmark that tests AI systems on realistic public authority case files – from case file analysis and drafting administrative decisions to preparing statements of defence. The focus is on legal quality, verifiable supporting evidence and the ability to recognise errors made by the authority itself.
Coming soonThe planned benchmark examines how AI systems work on contract documents across multiple negotiation rounds. It evaluates legal quality, adherence to client instructions and the handling of the counterparty’s proposed changes – directly from the resulting Word documents with tracked changes and comments.
Coming soonProf. Dr. Dr. h. c. Karl-Heinz Ladeur and Mansoor Koshan
11:10
Part of the webinar “KI im Unternehmen, aber wie? – KI-Richtlinie”.
Online webinar
Event details and registration at KI.SHMansoor Koshan
16:30–19:00
What legal questions arise as AI systems act increasingly autonomously – from responsibility and liability to the question of a possible legal status?
Part of “Vom Werkzeug zum Akteur – KI-Agenten und ihre Folgen für Wirtschaft, Recht und Gesellschaft”.
ZBW – Leibniz Information Centre for Economics, Düsternbrooker Weg 120, 24105 Kiel, Germany
Event details at ZBWMansoor Koshan
12:00–12:30
Part of the ARIC Remote Brown Bag Sessions.
Online webinar
Information about the event series at ARICMansoor Koshan
