News for hackers. Stuff that matters.

A comment on Addressing AI’s energy cost. Parent comment

Re:Addressing AI’s energy cost by mlyle on Wednesday June 08, 2022 @12:26AM (#31663180)
Likewise if one looks at total energy consumption in the US:

    2000: 98.7 quadrillion BTU / 281.4 million people = 350 million BTU/person/year
    2019: 100.5 quadrillion BTU / 328.3 million people = 306 million BTU/person/year.
On the other hand, increased use of global manufacturing and reliance on ocean transport is concealed by this metric. Still, it doesn't look like overall things are getting worse.

For the CO2 metric:

    2000: 20.9 tonnes per capita
    2019: 15.7 tonnes per capita
https://www.eia.gov/totalenergy/data/monthly/pdf/sec1_19.pdf
[ Parent | on HN ]
Re:Addressing AI’s energy cost by bmitc on Wednesday June 08, 2022 @01:27AM (#31663584)
According to the below EIA article, things are getting worse worldwide.

> Global electricity consumption continues to increase faster than world population, leading to an increase in the average amount of electricity consumed per person

https://www.eia.gov/todayinenergy/detail.php?id=44095#:~:tex....

[ Parent | on HN ]
Re:Addressing AI’s energy cost by mlyle on Wednesday June 08, 2022 @02:03AM (#31663825)
Well, sure. You have developing countries... developing, in addition to the (difficult to quantify) outsourcing effects that I mentioned above.

Note that I really want electricity usage to go up a lot, too. It's just overall primary energy usage that we want to stay level or go down. That's why I used the primary energy usage metric. After all, moving to an electric car or replacing natural gas heating with a heat pump "increases electricity usage."

[ Parent | on HN ]