Understanding legal AI and its potential

Today, we believe we are in the age of artificial intelligence (AI), a break-through that has grabbed our imagination in every possible way. But what if we said, the birth of AI took place in the mid-1950s?  

Yes, that’s right! AI was born as “Logic Theorist”, a ground-breaking programme engineered to mimic the problem-solving skills of a human being. Coined ‘artificial intelligence’ in 1956, of course at the time, few recognised its earth-shattering potential as the concept was met with skepticism and shrugs.  

Fast forward to 2024. AI is reshaping our world. From the smartphone in our pocket to self-driving autonomous vehicles to life-saving medical diagnoses and bleeding-edge scientific research, AI is pushing the boundaries of human capabilities, transforming industries, and redesigning  how we live and work. It is safe to say, the AI revolution is here. 

Different types of AI

While the technology’s latest incarnation – i.e., generative AI – has truly put AI in the spotlight, on a more practical level, AI-led automation has been around for a while in recent times.  We already use it in our daily lives. Just think Amazon and Google! 

There are different types of AI:

• Rule-based AI: This is one of the earliest forms of AI, where systems are programmed with a set of rules and logic to make decisions and solve problems. For example, based on specific rules, the system finds the most relevant and authoritative sources of information for a legal issue, ranking the results according to recency, relevance, and citation frequency. 
• Machine Learning: This involves using statistical and probabilistic techniques to train algorithms to learn from data, identify patterns, and make predictions or decisions without being explicitly programmed. 
• Deep Learning: A subset of machine learning, deep learning uses artificial neural networks inspired by the human brain to learn from vast amounts of data. Deep learning has proven highly effective for tasks like image and speech recognition, natural language processing, and computer vision. 
• Natural Language Processing (NLP): NLP focuses on enabling computers to understand, interpret, and generate human language. It involves tasks like speech recognition, text analysis, translation, and language generation. 
• Language Models: Language models are a type of artificial intelligence system that are trained on vast amounts of text data with the goal of understanding, generating, and analysing human language.

Some important things to know about language models:

• Training Data: Language models are typically trained on massive datasets of text from the internet, books, articles, and other sources. This allows them to learn patterns and relationships in how language is structured and used.
• Neural Networks: Most modern language models use deep learning neural network architectures, which enable them to process and make predictions about sequences of text.
• Natural Language Processing: Language models are a core component of natural language processing (NLP), allowing AI systems to perform tasks like text generation, translation, summarisation, question answering, and sentiment analysis.
• Large Language Models (LLMs): Advanced language models like Gemini, GPT, etc. produce contextualised word representations, meaning the same word can have different representations depending on the context it appears in.
• Applications: Language models power many commercial AI products and services including virtual assistants, chatbots, predictive text keyboards, content filters, code assistants, and more.
• Text Generation: One major use of language models is text generation - they can generate fluent, contextual text, based on an initial prompt or input.

LLMs are now widely referred to as ‘generative AI’. This innovation represents a key breakthrough for its ability to generate human-like natural language. However, as outlined above generative AI is only a subset of AI. 

• See the new iManage AI features to help streamline document and email management 

The generative AI bandwagon in the legal industry

Generative AI has grabbed attention across industries, and the legal sector is no exception. Not only are legal AI startups attracting venture capital funding, but acquisitions are also rife as legal tech vendors vie with each other to steal a march on their competition. The mere mention of AI in product roadmaps can now move share prices. 

Of course, it’s not a one-sided story. This heightened activity on the part of technology vendors is driven by users. A recent report highlights that 95% of legal professionals say generative AI will have a notable impact on law. Legal professionals are excited as the technology’s potential to make their lives easier is clear to see. 

At the same time, they are not wearing rose-tinted glasses. Generative AI is undoubtedly the most powerful technology in the legal sector thus far, but it’s not without challenges. 90% of lawyers acknowledge concerns about negative outcomes when using generative AI. For instance, trust in the output is a critical requirement, but issues such as accuracy, hallucinations, biases, and non-ethical advice put it at risk. 

To illustrate, a legal professional uses generative AI to identify a citation from a previous case to refer to in a legal document for a current client. How does this busy lawyer know that the AI tool hasn’t hallucinated to make up the citation? A Colorado lawyer who turned to AI for a civil litigation case, paid a rather heavy price (his job and reputation) on learning in court that the AI cited legal cases in his document didn’t exist. Similarly, how can a professional be 100% sure that a generative AI-generated summary of a 200-page document includes all the pertinent points? Or that an updated clause is truly an accurate and ‘good’ clause to use? The impact of inaccurate output can be damaging for legal professionals and clients alike. 

Additionally, ensuring client confidentiality is paramount. Ensuring that the use of AI stringently complies with the SRA’s Code of Conduct, the GDPR, and other country-specific data protection legislations – alongside the evolving AI regulations – is vital.  

AI use cases for legal 

The promise of AI is much broader than generative AI. The AI industry globally is expected to hit almost $830 billion value by 2030, nearly $90 billion more than the earlier forecast.

In legal, there are several use cases for professionals today that can remove the “drudgery” so that they can focus on the high-value, strategic aspects of their work. Over a quarter of legal professionals are now using generative AI tools at least once a month. Top priorities for using generative AI in the immediate future are legal drafting at 91%, researching legal matters at 90% alongside communication-based tasks at 73%. 

Here are some noteworthy use cases:  

• Email filing: AI is the perfect ‘mailbox assistant’ to help file emails in the right locations in the document management system (DMS). AI can read the emails in legal professionals' inboxes and learning from past behaviour, intelligently suggest which matter workspaces they should be filed. 
• Auto classification of documents: AI can review documents and automatically classify by type, tag, and apply metadata for more accurate search results and quicker retrieval. For instance, this capability can be useful for contract retrieval and remediation. Say, due to a change in regulation, certain clauses need remediation. Imagine the humungous task it would be for a legal team to manually wade through a corporation’s entire contract estate to find the contracts that might be affected by the changed regulation, and then appropriately remediate.  With AI assistance, the contracts can be found, analysed, and adjusted to accommodate the change in regulation. 
• Q&A with content: This is a good use case for generative AI. Legal professionals can ask the AI assistant questions about specific data points in documents. For example, “summarise X document including A, B, and C data points.”
• Knowledge management: AI combined with Conditional Forms can greatly simplify the knowledge submission, curation, and classification process in the knowledge management (KM) system for knowledge managers and legal professionals. AI can then automatically classify the documents in the relevant taxonomies.
• Closing sets: Once closing sets are added to the KM system, AI can assist legal professionals in classifying the various types of documents in the closing set. Legal professionals can use AI to identify important information such as how the deal was structured, and what creative clauses were drafted, for use in similar future deals. AI models can even review several different closing sets to, for example, determine which one would be the most suitable for an upcoming deal. 
• Know Your Customer (KYC) compliance: With AI, the automation within KYC and AML compliance processes can be further enhanced. Legal organisations can utilise natural language processing capabilities (chatbots) to navigate the labyrinth of in-house systems and processes.
• Threat identification: Machine learning techniques can be used to identify anomalous activity that could potentially signal malicious behavior.  
• Legal research: Legal professionals can use AI to conduct searches and ask legal questions in legal intelligence and content repositories. 

Future of AI in legal 

AI is going to be part of the fabric of every legal organisation. Major law firms and legal tech companies are actively researching and developing all manner of AI tools. By the end of 2024, all spending on legal tech AI tools and software is expected to reach about $37 billion globally. The technology’s potential is too hard to ignore. From a user standpoint, AI will augment the capability of legal professionals. At a business level, AI will change business models and drive new ways of working. The organisations that find the right balance of human expertise and ingenuity with AI capabilities that seamlessly align and scale, will be the winners.

The technology is very much in the early stages of its evolution. While AI presents intriguing opportunities in the legal sphere, it also raises knotty questions that the industry needs to carefully unpick, especially around risk, bias, discrimination, ethics, and compliance. Governance frameworks are evolving to ensure checks and balances as this technology advances. 

It’s an exciting space!

See also:

• New iManage AI features for legal professionals

• Legal AI: What's driving AI adoption

FAQs

How is AI used in the legal industry?

AI is used in a number of ways, from automatic email filing, knowledge management, and natural language processing-led assistance to meeting security, risk and compliance standards. 

How many law firms are using AI?

Research indicates that 26% of legal professionals use generative AI tools at least once a month. Respondents from academic institutions and large law firms were the most likely to use these tools for their work, at 33% and 32% respectively. 

How does generative AI affect the legal industry?

This technology is changing the legal industry and perhaps even the profession in a significant way. There is no getting away from the fact that the use of generative AI will only increase in the future, impacting everything from confidentiality and compliance, to communication, billing practices, ethics, and discrimination. All this impacts how law firms and corporate counsels ensure Academic institutions are determining ways of incorporating AI into teaching and assessment methods. Industry bodies are bringing out guidance to help professionals adopt the technology in a responsible manner. The Courts and Tribunals Judiciary has produced guidance to assist the judiciary, their clerks, and other support staff on the use of AI. Likewise, The Bar Council has issued guidance for barristers on how to navigate the growing use of large language models.