Vendors often use confusing or misleading terms when discussing the capabilities of their AI. Terms like “Data Capture” or “Data Extraction” give the impression that artificial intelligence can pull searchable, structured data directly from text and that’s simply not the case to a degree of accuracy that’s of any use for most legal use cases.
The text itself can be extracted. Letters and words can be gathered and grouped. But converting that text into high quality structured data that answers legal questions requires an understanding of the complexities inherent in contract prose as well as an understanding of how that information interacts with the text in related agreements. This is not a corner contracts AI has yet rounded for even seemingly some of the simplest terms.
A simple example
Let’s look at dates. They’re just numbers after all and we all know computers love the simplicity of counting. Effective dates are a critical identifier in finding an agreement, grouping it within a contact family, and using the resulting data efficiently. Well, similar to names, dates must be converted into a piece of data that can be grouped and analyzed in order to be understood.
However, the effective date could be defined as the commencement date or execution date. It might exist in the preamble or the signature block. It might be written clearly as a number but in either the US or UK convention. Any of these might cause comprehension errors for AI extraction or OCR search. The AI may pick up on your patterns after thousands of attempts at pattern recognition but stumble with insufficient date when introduced to client paper. These misunderstandings, which are so easy for AI to make, will ultimately result in data that grows to be distrusted by users.
While artificial intelligence is often pitched as a replacement for human analysis, at this point it can’t be trusted to solve even simple contract problems without some human support.
To learn more about how AI/ML is impacting the legal industry, read The Third Wave of Legal Technology Transformation: Changing the What of the Legal Practice.