Ep 38: Oscar Co-Founder Mario Schlosser on Lessons from Implementing LLMs and How AI Will Impact Healthcare
Vital Signs with Jacob Effron and Nikhil Krishnan
00:00
Limitations of Language Models in Customer Service
In this chapter, they discuss the limitations of language models (LLMs) and provide an example of how LLMs may not accurately summarize customer service phone calls. They explain that LLMs run out of layers and struggle with tasks that require processing and counting multiple tokens. They also mention the layer by layer processing of LLMs and how certain layers are responsible for identifying the final answer.
Transcript
Play full episode
