Today, exceptional communicator Lilith Bat-Leah explains why "Data-Centric ML Research" trumps our typical focus on model capability, with examples from her extensive Legal A.I. background.
Lilith:
Has over a decade of experience specializing in the application of ML to legal tech.
Is Senior Director of A.I. Labs at Epiq, a leading LegalTech firm that has over 6000 employees.
Has published work on evaluation methods for the use of ML in legal discovery as well as on Data-centric ML Research (DMLR).
Is co-chair of the DMLR working group MLCommons and has organized DMLR workshops at [ICML] Int'l Conference on Machine Learning and ICLR, two of the most important A.I. conferences.
Holds a degree from Northwestern University, in which she focused on statistics.
Today’s episode will appeal primarily to hands-on practitioners like data scientists, AI/ML engineers and software developers.
In today’s episode, Lilith details:
How A.I. is revolutionizing the legal industry by automating up to 80% of traditional discovery processes.
Why 'elusion' is a critical metric that only exists in LegalTech — and what it reveals about machine learning evaluation.
The surprising reason why we should stop obsessing over model improvements and focus on something that takes up 80% data scientists’ time instead.
How she grew from being a temp receptionist to an A.I. lab director by falling in love with statistics.
The SuperDataScience podcast is available on all major podcasting platforms, YouTube, and at SuperDataScience.com.
