Humans have a seemingly uncanny knack for solving complex coordination problems when communication and prior planning is limited by uncovering the shared cultural or knowledge-based default in a such situation, concerting the intentions and expectations and land on the above foci, named after economist and game-theorist Thomas Schelling.
Cooperative experiments demonstrate that a team of individuals acting towards shared end will pick the same time and place for a rendezvous. Part of the allure of AI models is that they seem also quite good at coordination problems—from predictive text, to routine emails to proofreading to peer-review, insofar as they have been trained on the social norms that we draw on as well to achieve a common goal. Artificial intelligence has a worse track record when it comes to something genuinely innovative or unprecedented, and moreover may erode the implicit social bargain that underpins cooperative efforts. The routine is also ritual and outsourcing them, like the above onerous tasks, dulls not only the refining practise when it comes to composing an email—which is also the author’s assessment of their audience—but of course lands as disingenuous and meritless when one can’t be bothered to dash off a good reference or buy someone a gift that was not generated by algorithm. What do you think? We’ve always been taking short-cuts but subverting ceremony altogether seems more serious. More from Henry Farrell at the link above.
