Do You Even Need That Trillion-Parameter Model?

August 5
15 mins

View Transcript

Episode Description

SUMMARY: We continue our Models and Money series. In this episode, Brian and Aaron explore the current state and future of AI models, focusing on model size, model harnessing, and intelligent model routing. They discuss whether bigger models are always better, the economics of AI, and how enterprise applications can benefit from tailored AI solutions.

SHOW: 1051

SHOW TRANSCRIPT: The Enterprise AI Show #1051 Transcript

SHOW VIDEO: https://youtu.be/tkJmeazn8Bs

SHOW SPONSORS:

Topic: When will the models/harnesses be good enough?

  • Why now? Benchmark Maxxing - cost to build/host/maintain 1+ trillion parameter model
  • Past: The “wow” moments in versions really stopped around GPT4… (maybe?)
  • Present: Race to the top/bottom, millions spent to gain SOTA for a few days
  • Future: Will the pendulum swing back? Will bigger/faster always rule?

FEEDBACK?

See all episodes