International Commercial Arbitration and Technology – An Authors’ Interview with Generative Artificial Intelligence

Piotr Wilinski, Maciej Durbas

Research output: Contribution to journalReview articleAcademicpeer-review

19 Downloads (Pure)

Abstract

It has been stated that generative artificial intelligence “is not yet accurate enough to provide legal information directly to laypeople.”Jinzhe Tan et al., ChatGPT as an Artificial Lawyer?, in AI4AJ 2023, Workshop on Artificial Intelligence for Access to Justice 7 (L. Karl Branting, ed., 2023), CEUR Workshop Proceedings, https://ceur-ws.org/Vol-3435/short2.pdf.X On the other hand, many sources label the introduction of generative AI as a milestone in the legal profession and international commercial arbitration.See, e.g., Claire Morel de Westgaver, Canvassing Views on AI in IA: The Rise of Machine Learning, Kluwer Arbitration Blog (July 12, 2023), https://arbitrationblog.kluwerarbitration.com/2023/07/12/canvassing-views-on-ai-in-ia-the-rise-of-machine-learning.X The authors of this paper took the task to conduct an “interview” with the generative AI to assess what it “thinks” about its newfound role in international commercial arbitration.

The quotation marks are deliberate. The “AI revolution” of 2022–2023 refers to the emergence of generative AI software, i.e., large language models (“LLM”) that could generate new data (in particular text) based on generative models.See Erik Brynjolfsson et al., Generative AI at Work 1 n.1 (Nat'l Bureau of Econ. Rsch., Working Paper No. 31161 2023).X The authors share the opinion that generative AI should not be compared with artificial general intelligence,See, e.g., Andrzej Kisielewicz, Bajki o sztucznej inteligencji i prawdziwe zagrożenia [Fairy tales of AI and real threats], Wszystko Co Najważniejsze (July 18, 2023), https://wszystkoconajwazniejsze.pl/andrzej-kisielewicz-bajki-o-sztucznej-inteligencji-i-prawdziwe-zagrozenia/.X i.e., software or machine that could accomplish any unspecific task that human beings could perform or even surpass human performance.Sam S. Adams et al., Mapping the Landscape of Human-Level Artificial General Intelligence, AI Mag., Spring 2012, at 25, 26.X Hence, one could not “talk” or “interview” generative AI software. Nor could such software share “its views” on a given subject. Accordingly, one could only input data to software (i.e., an interview question) and obtain other data in return (i.e., generative AI’s answer).
Original languageEnglish
Pages (from-to)46-118
JournalITA in Review
Volume5
Issue number3
Publication statusPublished - 2023

Fingerprint

Dive into the research topics of 'International Commercial Arbitration and Technology – An Authors’ Interview with Generative Artificial Intelligence'. Together they form a unique fingerprint.

Cite this