Marion Ranchet (@marionranchet): "Data inconsistency is already the norm in AI. Numbers thrown at us so far about the reach of LLM models: -> App downloads (says nothing about actual usage) -> Website and app visits (says nothing about actual users) -> Active users (what qualifies as one? what’s the subscri…"

Data inconsistency is already the norm in AI.

Numbers thrown at us so far about the reach of LLM models:

-> App downloads (says nothing about actual usage)

-> Website and app visits (says nothing about actual users)

-> Active users (what qualifies as one? what’s the subscriber count? Free vs Paid tier? B2C vs B2B users?

->Daily vs monthly vs total user data

Below is what I could gather.

What should be THE key performance metric by which we measure a model’s success?

Got more numbers? Come share them below.

————

This week in Streaming Made Easy, I analyzed so the distribution strategy of the models themselves, their reach and pricing.