Structured data – clean tables, standardized formats, defined fields – have been the foundation of data analysis across industries, including health care and litigation. Increasingly, however, valuable information lives outside those boundaries. Organizations today face growing volumes of unstructured data like documents, emails, and free-form text, which create new challenges in both scale and complexity.
But recent advances in large language models (LLMs) have expanded the breadth and depth of insights that can be extracted from this data. These tools allow us to analyze unstructured text with a level of speed, consistency, and depth that was previously out of reach.
Scroll through the visual below to learn more about the challenges of analyzing unstructured text.
These challenges emphasize the need for a flexible, context-aware approach for analyzing large volumes of text.
Using Generative AI to Extract Meaningful Information from Text
To meet this challenge, Analysis Group uses generative AI (GenAI) language processing capabilities to efficiently extract insights from large collections of text. We use pre-trained LLMs to better search, parse, and extract valuable data from free-form text.
GenAI has surged in popularity, especially LLM chatbots. These models are trained on enormous amounts of natural language and can generate text by predicting one word at a time, such as a plausible answer to a question.
After the user asks a question, the LLM breaks down the text into tokens – the small units of text it processes (sometimes full words, sometimes parts of words) – analyzes those tokens, and selects the text most likely to follow.
Because these GenAI models make educated guesses about the next word, they may confabulate (their guesses may be wrong) or give inconsistent answers (their answers may change over time). Firms have rushed to integrate this technology into their processes and offerings, but some may not understand its potential and limitations. Drawing on our deep expertise in traditional economic, statistical, and data-driven methodologies – alongside hands-on experience evaluating emerging GenAI tools – Analysis Group applies AI tools thoughtfully and selectively to litigation and consulting work, generating incremental value for clients.
Analysis Group leverages GenAI to obtain faster and better results, minimize key risks and pitfalls, and improve text analysis through both targeted use of third party tools and internally developed custom software. Additionally, we have internal capabilities to appropriately handle any sensitive data contained in textual sources.
Example use cases
The examples below showcase how AG used AI tools to increase efficiency and enhance search results in textual analyses. The gains enabled teams to reallocate effort toward more complex, judgment-intensive aspects of the analyses. These examples are illustrative and outcomes will vary depending on the nature of the data and the tasks at hand.
