The use of Generative Artificial Intelligence (GenAI) tools and, specifically, GenAI chatbots such as ChatGPT, Google’s Bard or Claude is becoming increasingly common in the legal industry. GenAI tools have the potential to enable increases in efficiency, accuracy and productivity across a number of legal tasks and activities. However, users must be aware of, and take into account, the risks and challenges which exist with these tools.
The Technology and Construction Solicitor’s Association (TECSA) is a leading Adjudicator Nominating Body in England and Wales. Adjudicators on its panel include Sir Rupert Jackson who was the judge in charge of the Technology and Construction Court and later Lord Justice of Appeal. TECSA has produced guidance for adjudicators who may use GenAI. TECSA encourages adjudicators to keep abreast of developments in GenAI, its potential benefits and uses, as well as its risks and limitations.
When adjudicators use GenAI, they must use it appropriately (either for decision writing or otherwise) and it is crucial for adjudicators to employ GenAI tools responsibly, ethically, securely and safely.
TECSA provides the following guidance for adjudicators [source www.tecsa.org.uk]:
‘Data Privacy and Security
7 Many GenAI tools openly available in the market (e.g. ChatGPT) are not private and confidential, even if they are a paid-for subscription. Accordingly:
7.1 As a general rule, do not put any data or documents, that are not already publicly available, into a GenAI tool.
7.2 Never upload sensitive or confidential information to a GenAI tool including, but not limited to, party names, addresses, project titles/details, financial information, or people’s names and details.
7.3 Protect the confidentiality and integrity of the parties and their data, either used in the decision-writing process or otherwise, at all times. Ensure that all communication and data exchange with these systems are secure and confidential.
7.4 Ensure that the GenAI tools comply with data privacy regulations, such as GDPR in the UK, by securely handling and storing sensitive information.
Responsibility, Supervision, Quality Control and Oversight
8 Adjudicators remain personally responsible for their Decisions, regardless of what tools are used in drafting and/or research (e.g. GenAI or otherwise).
8.1 Implement a rigorous oversight process to review and verify all AI-generated content, ensuring its accuracy and compliance with legal and regulatory standards.
8.2 Edit, develop and refine all AI-generated content to ensure appropriateness for the intended audience, avoid potential copyright infringement (if any), include a human touch and ensure the information is accurate, relevant and aligns with the specific case details.
8.3 Establish internal quality control procedures to review content generated by AI for clarity, accuracy, and consistency with legal standards.
8.4 Review service terms of any AI platforms used to understand indemnities, exclusions, liabilities, etc.
Understanding, Training & Continuous Improvement
9 Like any other drafting, research, or productivity tool, it is important to understand how the specific GenAI tool works.
9.1 Develop a deep understanding of the specific GenAI tool being used, including its algorithms and data sources.
9.2 Be mindful of the capabilities, limitations and potential biases of using GenAI tools in any activity (decision-writing or otherwise).
9.3 Keep up-to-date with any developments in the specific product, advancements in AI technology generally and best practice with regard to how GenAI may assist decision writing and other activities.
Human Expertise
10 GenAI is not at a point where it can fully replace human expertise and human experience. Indeed, some GenAI tools will provide inaccurate information, hallucinations and/or wrong answers at times.
10.1 Use GenAI as a supportive and enabling tool, rather than relying solely on automated content generation.
10.2 Utilise legal expertise and human judgment when reviewing, editing and finalising any AI[1]generated content.
10.3 Ensure that AI-generated content is accurate and aligns with the law and case-specific circumstances.
Transparency & Documentation
11 If and/or when AI is used for content generation in writing decisions, adjudicators should be transparent and open about its use.
11.1 Clearly communicate to all parties involved in the adjudication process when GenAI tools are proposed to be employed for content generation in decision writing. Ensure that the extent of AI involvement is transparent.
11.2 Provide an explanation of the GenAI’s role in decision drafting, emphasising that it is a tool to aid, not replace, the adjudicator’s judgment.
11.3 Maintain comprehensive records of the decision-writing process, including any use of GenAI tools, to ensure transparency and accountability.
Ethical Considerations & Bias Mitigation
12 When using GenAI tools, it is crucial to take into account ethical considerations to ensure fairness and prevent discrimination. For example, GenAI tools have the potential to create, replicate and/or amplify bias, depending on the data it has been trained on.
12.1 Be alert to risks like algorithmic bias. Implement measures to identify and mitigate biases in GenAI tools.
12.2 Regularly audit the output for fairness and take corrective actions as necessary.
12.3 Review AI-generated content critically to identify and rectify any potential biases that may inadvertently emerge.
12.4 Ensure that any use of GenAI aligns with ethical principles and professional standards governing adjudication which may be published from time to time.12.5 Uphold the principles of fairness, impartiality, and due process throughout the adjudication and decision-writing process, regardless of how or if GenAI is used securely handling and storing sensitive information.’