What is GenAI?
Generative Artificial Intelligence (GenAI) is a type of machine learning, such as ChatGPT, that uses unsupervised learning and generative models to produce new content, such as text, images, songs or other media. Large Language model GenAI is capable of taking raw data from any accessible source and “learning” to generate statistically probable outputs, based on common patterns and structures taken from its training data. Recent developments allow GenAI to be trained on billions of data points, with each iteration growing the systems' knowledge exponentially.
How are Lawyers using GenAI?
Artificial intelligence is not new to the legal profession. For example, since 2016/17 many large law firms have used Kira, which is a more limited form of AI that automates the extraction and analysis of contractual terms.
However, the use of modern GenAI is a recent development which fewer firms are actively using. The 2023 Thomson Reuters Institute report 'ChatGPT and Generative AI within Law Firms'[1] revealed that only 3% of law firms are already using GenAI for law firm operations, with only an additional 2% saying they are actively planning to use it. 60% said that they had no plans for GenAI use at all.
Two of the City firms who are using GenAI, both announced earlier this year their partnership with Harvey, a GenAI platform created specifically for legal work, including research, drafting and other routine legal tasks. Similarly, another large firm is using a platform called LUCY to automate part of its corporate-law transactions. These GenAI systems read results found across a law firm’s database to provide a summary to user questions. They do not search the web generally. Two firms have created their own internal GenAI chatbots, and a few others have all shown an interest in the CoCounsel platform recently bought by Thomson Reuters, which is a GenAI legal assistant.
DACB already uses AI within aspects of its own IT infrastructure and has recently set up an internal artificial intelligence Committee to prepare DACB’s own AI Strategy to remain at the forefront of digital innovation and transformation.
What are Clients and Businesses saying?
LexisNexis' 2023 report, 'Generative AI and the future of the legal profession'[2], revealed that almost half of in-house counsel respondents expressed that they expect their law firms to be using GenAI in the next 12 months, including 11% who expect firms to already be using the technology.
Only 8% didn’t want AI used on their work, although two-third of participants had mixed feelings about the impact of GenAI on the legal profession.
What are the Risks?
The most obvious is unchecked errors in GenAI output. There are other risks. Common concerns that are regularly raised about GenAI include:
1.Intellectual Property Risks – Who owns the input and output data?. With unfettered access to the web the training process for GenAI can run the risk of infringing the copyright, trade marks, patents and related rights of the owners of the training data. In addition is the output the original work of the user?
2.Ownership Clauses – AI vendors may use service agreement clauses to reuse input data to refine their systems or to retain ownership of the output data. Fundamentally it should be ensured that such clauses do not conflict with professional standards concerning the ownership and control of legal advice or compromise data confidentiality.
3.Data Protection – Unknowingly, personal data or confidential information may be included in the datasets that are used to train GenAI systems. Firms should be careful not to relinquish control of the data they process and store when involving GenAI systems in their work.
4. Cybersecurity – Using GenAI may increase system vulnerabilities to data breaches, and other malicious cyber activities, such as 'prompt injections' or data source corruption.
5. Output Integrity – When AI tools don’t have access to relevant data, they have been known to make up answers and information, known as ‘hallucinations’, which can lead to errors and inaccuracies. In such circumstances, a user would have no way of knowing when a system provides completely false information, unless they verify that information themselves.
6. Ethics and Bias – GenAI systems are only as impartial as the data they have been trained on. If data to train AI systems contains discriminatory material or bias, AI can spread this bias and lead to unjust and unethical outcomes and may impact ESG targets.
7. Reputation Risks – Reputational risks and brand damage may arise if GenAI results lead to negative consequences for clients, arising out of these risks.
The question arises as to whether liability policies will cover the use of unsupervised GenAI?
To assist law firms, The Law Society has recently published guidance on considerations law firms can take when implementing GenAI in their practice[3]. The SRA has also published its "Risk Outlook report: The use of artificial intelligence in the legal market" advising on the opportunities and risks GenAI poses, as well as offering guidance on how firms can manage such risks[4].
Conclusion
There is a clear expectation from many clients that firms should be taking advantage of GenAI in the near future, with many firms already making efforts to implement such tools. However, there are many areas of risk which need to first be addressed in order to ensure that the services provided to clients are not compromised. Whilst GenAI may have a bright future in the legal profession, it does not look like firms will be ready to part with their paralegals just yet.
[1] https://legal.thomsonreuters.com/blog/how-law-firms-can-use-ai-to-level-up-their-business/#how-many-law-firms-are-using-ai
[2] https://www.lexisnexis.co.uk/research-and-reports/generative-ai-and-the-legal-profession-report.html
[3] https://www.lawsociety.org.uk/topics/ai-and-lawtech/generative-ai-the-essentials
[4] https://www.sra.org.uk/sra/research-publications/artificial-intelligence-legal-market/
