The Silicon Valley Arbitration and Mediation Center is working towards publishing Guidelines on the Use of Artificial Intelligence (AI) in International Arbitration. George Agnew, litigation paralegal, provides an overview below.
Date : 31/10/23
The Guidelines seek to establish a set of general principle for the use of AI in arbitration. They are intended to guide rather than dictate. They do not intend to replace or override local AI laws of regulations.
The SVAMC Guidelines split into three chapters:
1) Guidelines applicable to all participants in international arbitration
2) Guidelines for parties and party representatives
3) Guidelines for arbitrators
Guideline 1: Understanding the uses, limitations and risks of AI applications
Participants should make reasonable efforts to understand the functionality, limitations and risks of the AI tools used in preparation for or during the course of an arbitration proceeding. This includes the following:
“Black-box” problem: Text produced by Generative AI is a product of complex probabilistic calculations rather than intelligible “reasoning”. AI lacks ability to explain their own algorithms. Where possible, participants should therefore use AI tools and applications that allow them to understand how a particular output was generated (“Explainable AI”).
AI tools may not be well-suited for tasks requiring specialised knowledge or case-specific information unless they are fine-tuned or provided with more relevant data.
Errors or “hallucinations”: This will occur when AI lacks information to provide an accurate response to a particular query. Errors can be reduced through “prompt engineering” and “retrieval-augmented generation”.
Augmentation of biases: biases may occur when the underrepresentation of certain groups of individuals is carried over to the training data used by the AI tool to make selections or assessments. Participants are urged to exercise extreme caution when using AI tools for this purpose.
Compliant: Using AI to conduct research on potential arbitrators or experts for a case Non-Compliant: Using it to select arbitrators or experts for a case without human input
Guideline 2: Safeguarding confidentiality
Need to ensure use of AI tools is consistent with obligations to safeguard confidential information. Confidential information should not be submitted to any AI tools without appropriate vetting and authorisation.
Participants should review the date use and retention policies offered by the relevant AI tools.
Compliant: Using AI for routine non-confidential tasks e.g., meeting scheduling or to research/summarise legal authorities in a third-party database. Non-Compliant: Submitting confidential information to a third-party AI tool as described above.
Guideline 3: Disclosure and protection of records
Some uses of AI by parties, experts and arbitrators may be uncontroversial and would not ordinarily warrant disclosure. There are certain circumstances where disclosing the use of AI tools may be warranted to preserve the integrity of the proceedings or the evidence.
A party seeking disclosure from another party should explain both why it believes that an AI tool was relied upon in the proceedings and how it would materially impact the proceedings and/or their outcome.
It is ultimately up to the parties and/or tribunal to specify the level of disclosure they want to institute for the proceedings.
Compliant: Using AI to generate document summaries for internal use or to identify and select the documents relevant and responsive to document production requests. Non-Compliant: Using AI to calculate damages without disclosing it. For an arbitrator to use AI to compare persuasiveness of parties’ submissions without disclosing it.
Guideline 4: Duty of competence or diligence in the use of AI
Parties and party representatives on record shall be deemed responsible for any uncorrected errors or inaccuracies in any output produced by an AI tool they use in an arbitration.
Compliant: Using AI to assist with drafting language for pleadings/written submissions or to assist in preparation of XX or find inconsistencies in witness statements.
Guideline 5: Respect for the integrity of the proceedings and the evidence
Parties, party representatives and experts shall not use any form of AI to falsify evidence, compromise the authenticity of evidence or otherwise mislead the arbitral tribunal and/or opposing party(ies).
Advancements in Generative AI and deep fakes can heighten the risks of manipulated or false evidence and can make it more costly or difficult to detect any such manipulation through forensic and other means.
Compliant: Using AI to produce demonstratives where the accuracy of the representation can be challenged by the opposing party by accessing the referenced source data.
Guideline 6: Non-delegation of decision-making responsibilities
An arbitrator shall not delegate any part of their personal mandate to any AI tool.
This Guideline does not forbid the use of AI tools by arbitrators as an aid to discharge their duty to personally analyse the facts, arguments, evidence and the law and issue a reasoned decision.
If an arbitrator used a Generative AI tool to assist in the analysis of the arguments or the drafting of a decision or award, the arbitrator cannot reproduce the AI’s output without making sure it adequately reflects the arbitrator’s personal and independent analysis of the issues and evidence at hand.
Compliant: Using AI to provide accurate summaries and citations to create a first draft of the procedural history of a case or generate timelines of key facts.
Guideline 7: Respect for due process
An arbitrator shall not rely on AI-generated information outside the record without making appropriate disclosure to the parties and allowing the parties to comment on it.
Where an AI tool cannot cite sources than can be independently verified, an arbitrator shall not assume that such sources exist or are characterised accurately by the AI tool.
Compliant: Using AI to distil or simplify technical concepts to come up with accurate and relevant questions for the hearing. Non-Compliant: Using AI to conduct independent research into substance of the dispute and base decision on such generated outputs without disclosing it to the parties.
Take The Next Step.
If you’re wondering about how we can help you focus more on outcomes, and worry less about hearing prep, book a tailored demo
or give us a call.