In this session, we will have a detailed look at the differences of the traditional machine learning models that have been used in the industry for more than a decade and the latest and greatest of generative AI.
We will explore the technical details of what impacts the model that we have used in the past and that we are using today and how the increasing complexity can impact our work in unexpected ways. We will discuss that the results are not only affected by the models themselves, but also by how we preprocess our data, how we encode it, what kind of inputs we use and what is the type of task that we have the models to execute.
And we will also explore why we have gone from being obsessed with statistics like recall and elusion rate to casually trusting our AI overlords, hoping that our common sense will be a sufficient guardrail.
But, most importantly, we will learn why we should not trust ducks!
Presenters
Jakub Weberschinke
eDiscovery Delivery Director, Actum Digital
Our Sponsors
