With an increasing emphasis on the use of large language models in legal practice, the Kelvin Large Language Model, known as KL3M, sets itself apart in two significant ways. Firstly, it is slated to be the first LLM developed from scratch specifically for the legal sector. Secondly, it receives the distinction of being the first LLM across all fields to be trained entirely on clean, legally permissible data, receiving a certificate of approval from the organization Fairly Trained.
The brains behind this distinguished endeavor are 273 Ventures, whereJillian Bommarito serves as the Chief Risk Officer. Bommarito played a key role in developing KL3M and the data set used for its training, overseeing the process of earning KL3M its Fairly Trained certification.
KL3M’s antique is no coincidence; it has been sculpted with the legal industry in mind. Bommarito reveals the intricate processes that led to the creation of this domain-specific LLM saturated with back-and-forth discussions on what the legal market stands to gain from KL3M and the circumstances under which a law firm would opt for this new model over other commercially available models.
Previously, Bommarito teamed up with Michael Bommarito and Daniel Katz, principals at 273 Ventures, to carry out an experimental test with GPT taking the bar exam. The trio, in conjunction with Katz’s wifeJessica Katz, were also founders of LexPredict, a legal AI and consulting firm that was acquired by global law company Elevate in 2018.
Bommarito’s comprehensive insights into the development and application of KL3M provide the legal industry with a profound understanding of this distinct large language model, its potential contributions to the market, and the dynamics of its utilization within the law firms.
More enthralling insights and detailed discussion with Jillian Bommarito on this significant development are available on the latest edition of LawNext.