Does AI benefit the world?

[We did this three-session activity three times during the year: first for cars, then the internet, then generative AI.]

[Global impact doesn’t lend itself to simple numerical representations, as what the numbers represent may not be evenly distributed.]

[Generative models are often discussed in terms of potential impact decades from now, so to start by measuring something that has already existed for decades helps ground discussion in less theoretical terms.]

[If we network homogeneously, our surrounding opinions will tend to conform similarly.]