8 Comments
Apr 14Liked by Michael Woudenberg

great piece! i have a couple of thoughts with a diff pov.

current llms are not only unreliable narrators (hallucinations et al) - they mostly regurgitate versions of what they’ve seen occur in statistically significant ways.

and we don’t yet know how to build models that can connect multiple sources of knowledge in a previously unseen manner.

given that - i would argue that the ai systems of today are only good at helping us automate workflows in a non-brittle fashion vs providing novel insight across domains.

Expand full comment
Apr 14Liked by Michael Woudenberg

It's nice to see a practical application of AI that isn't Hype or Panic. Well done!

Expand full comment
Apr 14Liked by Michael Woudenberg

I like it. I might augment my own personal definition of "polymath" to also include focusing on the juncture of two different fields of knowledge, although that's maybe more of an effect than a definition... but it's so fundamental to what I want from polymathic thinking that I feel like it belongs in my own personal definition. It's the goal.

Expand full comment
Apr 14Liked by Michael Woudenberg

I think if a person has the inherent curiosity of a polymath, then AI can help provide (or supplement) the requisite knowledge. Maybe the access to knowledge and a well prompted ai companion can help encourage Polymathic curiosity over time, but maybe not.

Expand full comment