Made me think about Uzzi and Spiro-- teams are not just multiple groups of people, but people embody mixtures of previous teams that they have been a part of.
I think that is one strength of these centrally trained models is that there is collective learning and transfer that is possible at scale, which can't happen with human teams. What is interesting is that you get the standard recommender system problem: increased individual diversity, low collective diversity. https://pubsonline.informs.org/doi/abs/10.1287/mnsc.1080.0974
Even if AI reduces the creative frictions within teams (or reduces most creative activity to solo+AI) it might have a net positive effect by amplifying "scenius" (scene genius) dynamics or how communities lead to phenomenal output. This might be especially true as AI amplifies exchange of tools and techniques so that scene members can more quickly remix and build on each other's innovations.
Hey friend! How's it going? Yes, potentially. Assuming that the members of a science are trying diverse things. But what if everyone is attempting the same things? Or that we get path dependence from remixing? .. Interesting question... just thinking through possible dynamics.
Made me think about Uzzi and Spiro-- teams are not just multiple groups of people, but people embody mixtures of previous teams that they have been a part of.
Nice. Very cool link.
I think that is one strength of these centrally trained models is that there is collective learning and transfer that is possible at scale, which can't happen with human teams. What is interesting is that you get the standard recommender system problem: increased individual diversity, low collective diversity. https://pubsonline.informs.org/doi/abs/10.1287/mnsc.1080.0974
Even if AI reduces the creative frictions within teams (or reduces most creative activity to solo+AI) it might have a net positive effect by amplifying "scenius" (scene genius) dynamics or how communities lead to phenomenal output. This might be especially true as AI amplifies exchange of tools and techniques so that scene members can more quickly remix and build on each other's innovations.
If you haven't come across the term, here's Kevin Kelly's overview of scenius: https://kk.org/thetechnium/scenius-or-comm/
Hey friend! How's it going? Yes, potentially. Assuming that the members of a science are trying diverse things. But what if everyone is attempting the same things? Or that we get path dependence from remixing? .. Interesting question... just thinking through possible dynamics.