The aim of this legal AI glossary is to help you understand and participate in the AI discussion. We are here to break it down for you, term by term. Consider this the only AI glossary you need.
Jump to ↓
AI architectures
Artificial intelligence (AI) · Agentic AI · Generative AI (GenAI) · Differences between agentic and GenAI · Professional-grade AI
Core AI technologies
Machine learning (ML) · Large language models (LLM) · Natural language processing (NLP) · Reinforcement learning
Data & information handling
Clean data · Data mining · Information retrieval · Structured and unstructured data
AI techniques
AI prompts · Retrieval augmented generation (RAG)
AI architectures: Conceptual types of AI systems
Artificial intelligence (AI)
Artificial intelligence is the field of simulating human intelligence in machines in order to enable them to execute tasks that otherwise require human intelligence. AI enables machines to learn using previous experience, adjust to new inputs, and perform tasks, exhibiting aspects of human cognition such as perception, reasoning, learning, and problem solving.
Agentic AI
Agentic AI is the next evolution of AI – capable of acting independently to accomplish complex, multi-step goals. Agentic AI can plan, reason, and execute multi-step processes following predefined objectives under human oversight and control.
Generative AI (GenAI)
GenAI refers to AI that can generate new content using large language models trained on proprietary data. GenAI requires assistance to create, such as descriptions of requirements and details, and needs fine-tuning in order to get the best results.
Differences between agentic and GenAI
While GenAI and agentic AI may look similar, they have differences and advantages to help with legal work. GenAI creates output by reaching to very specific input. Agentic AI takes it a step further by making decisions and taking action. Both types require human oversight to ensure accuracy and reliability of outputs.
For example, GenAI could draft a client proposal based on a standard template and recent case studies. Agentic AI could then submit the proposal through the firm’s CRM, schedule a follow-up meeting, and adjust the project pipeline to reflect the new opportunity.
Professional-grade AI
Professional-grade AI refers to advanced AI capable of delivering high-accuracy results, making it suitable for high-stakes professional environments, like legal work.
Professional-grade AI must be secure and private, developed by technology teams with deep AI expertise, and continually tested and authenticated by experts in the user’s field. There is a difference between professional-grade AI tools and current consumer applications, such as ChatGPT.
Core AI technologies: Fundamental technologies and models in AI development
Machine learning (ML)
Machine learning is a foundational aspect of AI focusing on the study of algorithms. It allows computers to understand data and relationships which enables them to perform certain tasks. Using multiple data points to identify patterns over time, machine learning powers technology to eventually make decisions or recommendations. This stands in contrast to traditional computers which required explicit instruction for every aspect of a task. For decades, machines had to be taught everything. With AI, they learn.
Machine learning in daily life
Your cellphone learns that you always head home around 5:15 pm. Based on that information, your phone is able to predict how long it’s going to take you to get home by analyzing factors such as time of day and actual movement of traffic on the day in question. It is learning from a combination of historical traffic patterns and from real-time data about that day’s traffic. Machine learning also shows up in personalized recommendations, face and voice recognition, and a host of other applications you see daily.
Large language models (LLM)
Large language models (LLMs) are a type of artificial intelligence algorithm that applies neural network techniques with lots of parameters to process and understand human languages or text using self-supervised learning techniques. Tasks like text generation, machine translation, summary writing, image generation from texts, machine coding, chat-bots, or Conversational AI are applications of the Large Language Model.
LLMs are highly relevant and important to lawyers when it comes to using generative AI (GenAI). It can help with crafting effective contracts, summarizing information, diving into legal research, and other legal work demands to safeguard all parties’ interests and meet legal standards.
Thanks to its use of LLMs, AI can analyze large volumes of precedents and current case law to provide clauses and language that have been judicially tested, thereby reducing the risk of future disputes.
Natural language processing (NLP)
Natural language processing focuses on understanding human language – both spoken and written – not robotic speech or restrictive text. Natural language processing applies algorithms to extract and analyze language data in a way that computers can process.
For machines to be able to process enormous amounts of data – to be able to mine it and organize it and ultimately, to understand and translate it – is imperative.
Natural language processing in daily life
Natural language processing means that when you’re searching for a new gym by searching for “gym,” your results will include most places that are focused on fitness, regardless of whether the name of the business actually includes the word “gym.” From a traditional gym, a CrossFit or yoga studio, Google understands that “gym” and “studio” in this instance have a similar meaning.
It also means that when you’re searching through court decisions, dockets and briefs, you don’t always have to use exact-match language to find precisely what you’re looking for.
Reinforcement learning
Reinforcement learning is a type of machine learning where an agent learns to make decisions by interacting with an environment. The agent receives feedback in the form of rewards or penalties based on its actions, and it uses this feedback to improve its decision-making over time. This process involves trial and error, where the agent explores different actions and learns from the outcomes to maximize its cumulative reward.
Reinforcement learning is particularly relevant to lawyers because it can be used to develop AI systems that assist with complex legal tasks. For example, reinforcement learning can be applied to legal text summarization, where an AI agent learns to generate concise summaries of legal documents by receiving feedback on the quality of its summaries. This can help lawyers quickly understand large volumes of legal texts and make informed decisions. Additionally, reinforcement learning can be used to optimize legal research processes, automate repetitive tasks, and improve the accuracy of AI-generated legal content.
Data & information handling: Data-related tasks essential to AI performance
Clean data
All data is not created equal. Clean data is properly maintained. In other words, incorrect, incomplete, or otherwise “bad” data is modified or removed. Redundant data, whether it’s unstructured or because the data is being pulled from too many sources, can also skew results. Taking a proactive approach to data cleansing can be a time-consuming, expensive endeavor. But anyone who has spent hours sending out holiday cards only to receive dozens in the mail due to outdated addresses can tell you, it is a critical component to creating trustworthy data.
Taking it a step further, clean data can also be the result of changing the processes that go into creating data in the first place. That discipline of managing data processes and ensuring data hygiene is called data governance.
Data mining
Data mining is a process of looking for relationships, correlations, and patterns within large data sets. Technology systems scour data and recognize anomalies within the data at a scale that would be impossible for humans. This analysis helps predict outcomes, finds potential wrongdoings, and notices questionable trends–and that information derived can be useful in a variety of ways.
Data mining in daily life
By analyzing the patterns of people who also buy or are interested in the same products as you, a store can make relevant suggestions based on that data. This same concept plays out in Netflix recommendations or the seemingly endless stream of targeted advertisements on Facebook and other social media channels.
Information retrieval
Information retrieval uses stored data to help you find what you want, when you want it. A web search engine like Google is a commonly used information retrieval system. Stored information is searched based on the words or phrases used in the query, and matched to the existing index of websites, data, content, and metadata that populates the web. It’s a lot to search through, but AI makes information retrieval fast and feasible for users across the globe.
Information retrieval in daily life
Alexa can tell you what time [insert your Favorite Restaurant] opens.
Your inquiry – Alexa, what time does [Favorite Restaurant] open – is the first step in the retrieval process. Alexa combs through information stored on the internet to find your specified restaurant, specifically the one located close to your current location, and assesses what piece of data on their business listing are the operating hours. The device must also be aware of the current day and time to ensure the information retrieved is a valid answer to your question.
Structured and unstructured data
To put it simply, structured data is organized data. It may be referred to as quantitative data. It is objective and easy to export to and store in Microsoft Excel. The way it is organized is consistent and easily identifiable, which makes data mining better. Structured data is also less complicated to analyze and distill.
On the other hand, unstructured data isn’t so easily exported, stored, or organized. And it’s the bulk of what most organizations deal with daily. It includes most text-heavy data, such as reports, Microsoft Word documents, emails, and webpages.
AI techniques: Methods or practices in using AI
AI prompts
AI prompts are inputs or queries that the user gives to an LLM AI Model, in order to get a specific response from the model. It can be a question, code syntax, or any combination of text and code. Depending upon the prompt, the model returns the response.
AI prompts are highly relevant and important to lawyers for several reasons:
- Accelerated research: AI can sift through statutes, case law, and regulations quickly, giving lawyers time to address more complicated issues.
- Automating repetitive tasks: Law firms can automate billing and document sorting with the help of AI.
- Increased business: With AI assistants, law firms can offer immediate answers to inquiries.
- Compliance monitoring: As changes in regulations happen, AI can provide instant updates to help keep law firms current.
- Drafting contracts: AI can assist in creating contracts that involve AI tools.
- Legal research: AI tools can summarize cases, identify precedents, and highlight key risks within minutes, which is especially useful when working under tight deadlines or reviewing complex legal documents.
Retrieval augmented generation (RAG)
Retrieval-augmented generation is a technique that enhances large language models (LLMs) by incorporating an information-retrieval mechanism. This allows models to access and utilize additional data beyond their original training set. RAG improves the accuracy and reliability of AI-generated text by retrieving relevant documents and using them as inputs to the AI before generating a response.
RAG is particularly relevant to lawyers because it allows AI tools to access and incorporate domain-specific information from legal documents, contracts, cases, and other primary and secondary sources. This enables a rich level of nuance and expertise for specialized fields such as law. By using RAG, legal professionals can benefit from more accurate and reliable AI-generated text, reducing the chances of model hallucination and ensuring that the information provided is grounded in authoritative sources.
Ready to engage, ready to lead
As AI continues to reshape the practice of law, having this knowledge base positions you at the forefront of legal innovation rather than scrambling to catch up. Tools like CoCounsel are already demonstrating how AI can augment legal work, helping legal professionals draft documents, conduct research, and analyze contracts with unprecedented efficiency and accuracy.
Whether you’re evaluating new AI solutions for your firm or simply engaging in more substantive discussions about the future of law, you’re ready to be a part of the evolution.
CoCounsel
Bringing together generative AI, trusted content and expert insights
Meet your AI assistant ↗
