Live webinar
·
November 10, 2026
·
10am PT / 1pm ET

The IP Turing Test: Measuring Quality in AI-Assisted Patent Drafting

Berkeley's Bo Heiden, LegalBench creator Neel Guha and former EPO Chief Economist Nikolaus Thumm on what the data actually says about AI-drafted patent claims, and what in-house IP teams can take from it.

Thanks. The recording is below, and a copy of the link is on its way to your inbox.

Meet the panel

Bo Heiden
Bo Heiden
Founding Faculty Director, Open Innovation Labs
·
UC Berkeley Haas

Bowman (Bo) Heiden is Founding Faculty Director of Open Innovation Labs at UC Berkeley's Haas School of Business and Executive Director of the Tusher Strategic Initiative for Technology Leadership. He co-directs the Center for Intellectual Property (CIP) at the University of Gothenburg and co-chairs the Technology, Innovation, and Intellectual Property program at NYU Law's Classical Liberal Institute. He is currently leading a team of Berkeley researchers evaluating the quality of AI-drafted patent claims against attorney work product.

Neel Guha
Neel Guha
Associate Professor of Law
·
Columbia Law School

Neel Guha is Associate Professor of Law at Columbia Law School, where his work sits at the intersection of law and machine learning. He created LegalBench, the leading benchmark for measuring legal reasoning in large language models, and his research on how to evaluate AI-generated legal work has been widely adopted across the legal technology industry. He holds a J.D. from Stanford Law School and a Ph.D. in computer science from Stanford.

Nikolaus Thumm
Nikolaus Thumm
Former Chief Economist, European Patent Office
·
Berkeley Fellowship Program, UC Berkeley Law

Nikolaus Thumm is an economist who served as Chief Economist of the European Patent Office and later as Senior Fellow at the European Commission's Joint Research Centre, where he worked on standard-essential patents, FRAND licensing and the economics of the patent system. He advises national and international organizations including the OECD, is an associate at TU Berlin, and is a fellow in UC Berkeley Law's Berkeley Fellowship Program, where his current research looks at how AI is reshaping invention and the future of the patent system.

Moderated by
Alec Sorensen
Alec Sorensen
Founder and CEO
·
Tradespace

Founder and CEO of Tradespace, the AI-native IP platform and law practice used by university technology transfer offices and corporate IP teams. Has spent his career in IP, from technology transfer to patent prosecution strategy.

About this session

AI is now everywhere in legal drafting, but there is still a fierce debate over the quality of AI compared to attorney work product. Both sides lack any real data or benchmarks to evaluate the quality of AI applications, or even attorney-drafted claims for that matter.

This is particularly concerning given the changes shaping the IP industry. 35% of attorneys report using AI to draft claims, and the emergence of AI-native IP law firms is forcing in-house teams to consider new models for prep and prosecution.

We're sitting down with three of the leading voices on AI and quality in law. Bo Heiden, Founding Faculty Director of Open Innovation Labs at UC Berkeley Haas and Co-Director of the Center for Intellectual Property, has been leading a team of Berkeley researchers evaluating the quality of AI-drafted patent claims. Neel Guha, Associate Professor of Law at Columbia Law School, developed LegalBench, the leading benchmark for AI-generated legal work; in his blind studies, professors at top law schools consistently ranked AI answers to exam questions above their colleagues'. Nikolaus Thumm, former Chief Economist of the European Patent Office and a fellow in UC Berkeley Law's Berkeley Fellowship Program, is studying how AI is reshaping invention and the patent system itself.

Tradespace CEO Alec Sorensen will moderate a conversation about what they've learned judging quality in a judgment-driven field, and what in-house IP teams can take from it as they weigh AI-native law firms and other approaches to patent prep and prosecution.

Can't make it live? Register anyway. Everyone who registers gets the recording afterward.

What you'll walk away with

  • What the data says. Early results from Berkeley's evaluation of AI-drafted patent claims against attorney drafts.
  • How to measure judgment. What LegalBench shows about benchmarking legal work that has no single right answer.
  • For universities. What this means for a tech transfer office filing provisionals and choosing counsel.
  • For corporate IP. How to weigh AI-native law firms and other models for patent prep and prosecution.
  • Your questions. Submit one when you register and the panel will take as many as they can live.

Good to know

Is it free?

Yes.

What if I can't attend live?

Register anyway. You'll get the recording once the session ends.

How do I join?

After you register you'll get a personal Zoom link and a calendar invite, plus reminders a week, a day and an hour before we start.

Can I invite colleagues?

Please do. Join links are personal, so send them to this page to register rather than forwarding yours.