The funny thing is, with AI each individual token is surprisingly efficient, but each query is burning 10s or 100s of tokens, and a single interaction can lead to 10s or 100s of queries. Factor in that there’s forced AI integrations into things that don’t need it on top of the millions of active users, the near constant training of new models, and suddenly its ballooned into an amount of energy that’s noticable on a global scale
Now try saying this about AI
Using the words thank you to respond to Alexa uses the same amount of gasoline a wood chipper takes to consume eleven spotted owls.
Unleaded, though!
The funny thing is, with AI each individual token is surprisingly efficient, but each query is burning 10s or 100s of tokens, and a single interaction can lead to 10s or 100s of queries. Factor in that there’s forced AI integrations into things that don’t need it on top of the millions of active users, the near constant training of new models, and suddenly its ballooned into an amount of energy that’s noticable on a global scale