By Jitendra Gupta, Head of AI and Data Science, Wolters Kluwer ELM Solutions
For a technology that was first described in 1935, AI is certainly misunderstood.
Perhaps you didn’t realize AI was that old, but it’s true. Mathematician and computer scientist Alan Turing first coined the concepts of AI and machine learning (ML) several decades ago. Turing theorized a machine that could “read what it finds,” allowing it to “modify or improve its own program,” according to Britannica.
So, why the continued confusion over what is essentially a legacy technology? Perhaps it’s because AI has always been a somewhat nebulous concept. How does a machine know what to modify or improve if it doesn’t think? How can we trust it to come to the correct conclusions? As corporate legal departments (CLDs) continue to ponder these questions, the advent of generative AI has made discovering the answers even more urgent.
But what are the differences between traditional AI and generative AI, and what are their use cases for corporate legal?
The evolution of traditional AI and the introduction of generative AI
One of the initial uses of AI in legal was in eDiscovery, but within the last decade, AI’s use has expanded as organizations started to realize they could use AI to do a lot of the jobs that humans would normally spend hours performing. These include managing claims, evidence discovery, reviewing billing, and other common and time-consuming legal tasks.
Since then, AI has continued to evolve. Now, it can improve billing guideline compliance and streamline bill review workflows. It can also use historical performance and other factors to help legal operations professionals estimate budget and cycle times (though not, of course, without some level of human intervention).
The introduction of ChatGPT in 2022 brought to market a form of generative AI that uses large language models (LLMs) to generate text. Where traditional AI completes a task, generative AI creates content that can be read and understood by humans. Almost immediately, legal professionals began to investigate what generative AI could create for them.
Two years later, the investigation continues. While it’s undeniable that generative AI has serious potential for the legal field, many corporate legal teams remain skeptical about the reliability of the content and are confused about how generative AI differs from traditional AI. Thus, they struggle with how to implement all types of AI in practical and useful ways.
Here’s a list of some of the most popular applications of traditional and generative AI:
| Traditional AI | Generative AI |
| Classification | Summarization |
| Prediction | Generation of text and code |
| Anomaly detection | Personalization of content |
| Natural language processing | Natural language querying |
| Creation of insights | Language translation |
| Metadata extraction | Formless applications |
| Clustering | Virtual assistance (i.e. chatbots) |
Traditional AI vs. generative AI: differences and legal use cases
Understanding the differences between traditional AI and generative AI can make it easier for corporate legal teams to discern when to use one or the other. Let’s break things down with a simple comparison of the two technologies to further that understanding.Think of traditional AI as a classification technology. It’s great at grouping items into categories—sets of charges on legal invoices, for example. Using ML and large amounts of data, it can derive certain inferences from these classifications, like using the sets of charges to tell if a law firm tends to deviate from established billing guidelines. Over time, AI can use the information it collects to provide intelligence that helps CLDs choose firms, establish appropriate budgets, and more.
Generative AI is a summarization and response technology. While AI groups outside counsel charges into categories, generative AI summarizes what the categorization means to the CLD. Using natural language processing (NLP), generative AI can also decipher queries (e.g., “How long will it take a matter of this size and complexity to be resolved?”) and deliver detailed responses.
Here’s a breakdown of current traditional and generative AI use cases in corporate legal:
| Traditional AI | Generative AI |
| Detecting anomalies in invoices, billing guidelines, etc. | Quickly creating standard contracts |
| Automating time-consuming processes, like bill review, firm selection, etc. | Delivering concise yet informative answers to users’ queries |
| Extracting metadata, which allows the technology to interpret datasets, extract meaningful information, and deliver accurate recommendations | Instantly summarizing content (for example, details of matters, billing guidelines, etc.) |
| Automatic enforcement of invoice pricing, billing guidelines, etc. | Personalizing content (particularly useful for legal engagement) |
What’s next for AI in legal?
AI is changing at a pace unlike most other technologies, and what was true last week may not be true right now. So, while it’s become somewhat of a fool’s errand to predict what will happen next, there are some educated guesses we can make with a fair amount of confidence.
First, while generative AI will certainly have a place in all CLDs, the extent to which organizations use the technology will largely depend on its reliability and usefulness. LLMs used by applications like ChatGPT are made more impactful when organizations insert their own data into the models, but that information is made accessible to the entire world. Using domain-specific small language models (SLMs) is a more targeted and perhaps safer option. SLMs allow corporate legal departments and law firms to train their own data within a small pool of domain-specific information, resulting in highly targeted and relevant results without exposing potentially proprietary information to the Internet.
Second, any remaining concerns about whether traditional or generative AI will replace corporate attorneys are likely to dissipate as users understand that AI cannot replace lawyers. AI is great for making recommendations, but it can’t replace the intuition and experience of a human being. Rather, it is, and will continue to evolve into, a digital helper that exists to make lawyers’ work easier, more productive, and more accurate.
Finally, AI innovation will continue apace. CLDs will be the beneficiaries of advancements in AI-driven legal bill review, firm selection, budget estimation, and other uses.
The rest, as they say, is anyone’s guess. As Alan Turing once said, “Machines take me by surprise with great frequency.” What surprises does AI have in store for the legal industry? Only time will tell.
