- New technology, such as artificial intelligence (‘AI’), is driving change in society and business. AI’s development, from theory to application, impacts not just legal practice but also the ethical duties of lawyers.
- Lawyers need to be able to use AI tools to be able to discharge their duties of competence and act in the best interest of clients. AI tools may also facilitate access to justice.
- AI also creates risks when it comes to maintaining confidentiality, legal professional privilege and independence.
The adoption of artificial intelligence (‘AI’) in society and as a tool for the practice of law impacts lawyers’ ethical duties. We look closely at five of these duties:
- Duty to act in the best interests of the client
- Confidentiality and legal professional privilege
- Access to justice
Rule 4 in the Legal Profession Uniform Law Australian Solicitors’ Conduct Rules 2015 (‘Solicitors’ Conduct Rules’) provides that a solicitor must ‘deliver legal services competently, diligently and as promptly as reasonably possible’. Competence may be defined as having knowledge of the law and being able to use the law (both substance and procedure) with skill to solve problems. It refers to technical proficiency. It also involves effectiveness and efficiency – the rule refers to diligence and promptness – which includes achieving an outcome in a timely and cost effective manner.
The impact of technology on the functioning of law and the legal profession was highlighted by the American Bar Association (‘ABA’) when it approved changes to its Model Rules of Professional Conduct. A comment was added to rule 1.1 (Competence) to make clear that lawyers have a duty to be competent not only in the law and its practice, but also in technology. A majority of US States have adopted the comment (Comment 8) which provides:
‘To maintain the requisite knowledge and skill, a lawyer should keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology, engage in continuing study and education and comply with all continuing legal education requirements to which the lawyer is subject.’
In jurisdictions outside the US where there is no specific rule or commentary addressing technology, such as NSW, the general requirement of competence may nonetheless be viewed as including competence with technology. Just as lawyers need to keep up with changes in the law, lawyers need to keep up with changes in technology. This argument can be made on the basis that it is not possible to carry out competent legal representation without being able to use relevant technology.
A straightforward example is legal research before and after the rise of electronic databases, the internet and various forms of search – keywords, connectors (i.e. AND, OR) and natural language. The way a lawyer finds the law has changed and with it the way a lawyer must think about finding the law. The lawyer who relies only on the textbooks, legislation and case law (including citators) physically in their office cannot be as comprehensive as the lawyer with access to online databases. However, the effectiveness of using those online databases requires knowledge of how to search.
Another common example involving AI is the use of technology assisted review (‘TAR’), using machine learning, in discovery or due diligence. Research studies have found it to be more accurate and less expensive than human review for large scale document review. TAR is a highly technical exercise. It involves an array of methodological choices, such as seed set selection strategies, choices among ‘learning protocols’ and evaluation of performance metrics. The use of TAR in discovery has been addressed by the Supreme Court of Victoria. The Practice Note on technology in civil litigation expressly adopts the following assumptions: (i) communications and dealings in modern society are predominantly conducted electronically and (ii) a large number of discoverable documents are stored by parties electronically. The Practice Note goes on to make clear that ‘the use of common technologies is a core skill for lawyers and a basic component of all legal practice’.
What does competence mean in relation to technology? It is the lawyer being able to:
- choose technology that is fit for purpose;
- use the technology correctly, including understanding its outputs;
- understand the risks associated with technology;
- challenge or interrogate technology (e.g. AI tools such as Robodebt or AI used for sentencing criminal offenders).
Importantly – lawyers can get help. They can retain experts, such as third party providers of TAR, to assist them in using relevant technology to provide representation. They can undertake training, including as part of continuing professional development, to learn how technology functions as well as its limitations.
Duty to the client
Rule 4 of the Solicitors’ Conduct Rules also requires that a solicitor ‘act in the best interests of a client in any matter in which the solicitor represents the client’. The duty to the client is typically characterised by requirements of loyalty, partisanship and acting in the client’s best interests. The lawyer is required to put the client’s interests before their own, although the lawyer is permitted to charge a fee for the services they provide.
AI may improve the quality and/or efficiency of the lawyer’s work. This can mean a better result for the client. AI can assist the lawyer in promoting the client’s interests by bringing about the outcome the client seeks – resolving a dispute or bringing a transaction to successful completion. For example, if TAR aids in finding the key documents in litigation or as part of a due diligence, then acting in the client’s best interests may require the use of the AI tool. AI can also save significant time so that lawyers may be found to be acting in an unethical manner where they bill for doing these tasks manually or using (for example) inferior technology. Failure to use technology may result in overcharging.
Put in the converse – it is not in the client’s interest to conduct representation in a way that is incomplete or takes more time and cost because the lawyer cannot use readily available technologies.
Confidentiality and Legal Professional Privilege
Rule 9 of the Solicitors’ Conduct Rules states that a solicitor must not disclose any information which is confidential to a client and acquired by the solicitor during the client’s engagement, subject to specified exceptions.
Relatedly, legal professional privilege protects from disclosure communications between a client and his or her lawyer, made for the dominant purpose of giving or obtaining legal advice or the provision of legal services, including representation in legal proceedings.
The issue of technology and confidentiality has been considered in a number of contexts, namely the use of mobile phones, email, wireless technologies (such as public wifi) and data storage in the ‘cloud’. It also arises in relation to AI.
The issue in relation to AI, notably machine learning, is that it needs data to be able to function. Client (or law firm) communications, including electronic versions of documents, may need to be analysed by the program as part of its training.
Machine learning applications generally require a high volume of data in order to be trained, so the status of this data must be clarified in order to maintain its confidentiality. For example, an AI product can be used to cite check a memorandum of law or court submissions to ensure that cases have not been over-ruled or questioned, and to suggest additional cases. The draft advice or submissions are confidential and privileged. However, for the AI product to perform its function, the draft and confidential memo/submission must be accessed by the program.
Another example would be the use of natural language processing in relation to speech. The AI product will need to be given the recorded speech to convert it to written text. This then raises a number of questions: Are the recordings being saved? If so, who owns them or has access to them? Are the recordings being used for AI improvement? What security measures are in place? If the recording contains confidential or privileged information, the lawyer needs to obtain assurances that the recording will remain confidential and not be disseminated. Indeed, it would be prudent to go further and require that the recording be deleted after it has been converted to text.
A lawyer may be tempted to try and avoid technology because of confidentiality concerns. However, this has to be weighed with the duty of competence and the duty to the client.
Independence is the ability to act and to exercise judgment free from external pressure. Professional independence is typically discussed in relation to different pressures upon lawyers – commercial interests, pressure from more senior colleagues or employers, and from clients. Independence can mean independence from a client seeking to influence advice, i.e. maintaining a detached or objective state of mind. At other times it can mean independence from the pressures and influences of others who might compromise lawyers’ loyalty to clients.
Rule 4.1.4 of the Solicitors’ Conduct Rules specifies that solicitors must ‘avoid any compromise to their integrity and professional independence’. Moreover in relation to advocacy, rule 17.1 states that the lawyer ‘must not act as the mere mouthpiece of the client’.
The first challenge of regulating lawyers’ use of AI is whether such ‘outsourcing’, even if on a small scale, could compromise a lawyer’s independent judgment. Arguably, if lawyers are overly reliant on an AI program – e.g. to tell them about precedent cases, to search for relevant documents in discovery, or even to predict the outcome of litigation
– they are not exercising independent professional judgment. In response, it is suggested that lawyers should supervise AI systems as they would their junior legal staff. Indeed this is the approach in the US in relation to complying with the competence requirement as discussed above. Yet this is complicated by the fact that lawyers may not be able to independently evaluate the functioning of the software due to their own lack of technical knowledge, which is quite different to supervising a junior lawyer. There is also the lack of transparency in a system’s functioning, or the fact that no one (even its developers) may fully understand how its outputs were generated. It may be difficult or even impossible for lawyers to assess how accurate the technology is. However, not all AI is unable to be tested. The supervised machine learning used in TAR is able to be tested using statistical sampling.
Access to justice
A lawyer has an ethical duty to facilitate access to justice. As the main possessors of legal knowledge and skills, and with a monopoly on exercising these, lawyers have a key role to play in ensuring access to the justice system, commonly through pro bono representation. The Solicitors’ Conduct Rules do not address access to justice expressly, but it is nonetheless recognised as an ethical duty. As Street CJ stated in Re Foster (1950) 50 SR (NSW) 149 at 151: ‘in a profession pecuniary success is not the only goal. Service is the ideal’.
The duty to facilitate access to justice may be met through AI. At the individual lawyer level and at the profession level there is scope for AI systems to be used to assist those that cannot afford legal services to obtain advice, the drafting of documents and dispute resolution services. An AI product once created can be used repeatedly without additional cost or effort. Expert systems and chatbots have been deployed to provide basic legal information and to assist those with disputes to commence proceedings and attempt some form of alternative dispute resolution.
However, it is also likely that lawyers will still have a role to play in supporting the person during or after the use of the AI-provided service or product, depending on the complexity of the problem or what is at stake. AI products might also help lawyers to work more efficiently and reduce their fees.
It is arguably no longer a question about whether lawyers should use technology but rather exactly what technology they should choose and the degree of knowledge they should have. Lawyers do not need to be experts in AI but adherence to the Solicitors’ Conduct Rules requires them to at least know what they do not know and take steps to address a lack of knowledge.
* This article is based on research for the authors’ book Artificial Intelligence and the Legal Profession (Hart, 2020).