What Doug Belshaw is describing here is what I've called 'regression to the bland'. When AI reduces a large body of things to what it thinks you will want to see (no matter how well founded) it tends to exclude "are unusual or fit a different profile than you were expecting are quietly removed from view." I experienced this when I was using Feedly's AI Leo to narrow down RSS results. The results were nothing unusual or surprising, which is not what I want when creating a newsletter like OLDaily. Belshaw also makes the useful distinction between choice - where we select from future options - and agency - where we actually shape these options.
Today: Total: [Direct link] [Share]
