This is a great paper by Mitsuru Igami framing breakthrough AI algorithms in the familiar logic of structural economic model estimation. doi.org/10.1093/ectj... But by implication, if structural models behave like board games, such as Chess and Go, how high can their degree of realism really be? - ThreadSky

About ThreadSky

p-hunermund.com • 8 days ago

This is a great paper by Mitsuru Igami framing breakthrough AI algorithms in the familiar logic of structural economic model estimation. https://doi.org/10.1093/ectj/utaa005

But by implication, if structural models behave like board games, such as Chess and Go, how high can their degree of realism really be?

Comments

jonthawk.bsky.social•8 days ago

I think the board game applications were just attractive proving-grounds for the AI algorithms, and that they are more widely applicable.

AlphaStar is a similar algorithm to AlphaZero but plays StarCraft at a high level. That seems complex enough for interesting economic applications.

plsecon.bsky.social•8 days ago

Isn't this expected? If we have a rational agent dynamic model, all the agent does is intertemporal optimization of an objective function. If we replace utility with 'points' and optimization with strategic play, we get a game in the non economic sense.

chrisadamsecon.bsky.social•8 days ago

What is the question? As I understand it, we are talking about methods for approximating optimal choice with finite time and computing power.

p-hunermund.com•8 days ago

Yes, but we're dealing with a state space that's maximally 10^171.

chrisadamsecon.bsky.social•8 days ago

Are saying that is small relative to modeling starbucks entry?

vegasgello.bsky.social•8 days ago

None shall pass...

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply