Speaker
Description
Since 2020, Microsoft’s emissions have risen by more than 30%, and Google’s by almost 50%. The reason, according to both companies, is the insatiable energy demand of generative AI. As a result, both companies are now openly wavering on their previous corporate emissions targets and making it clear that AI growth will take priority over achieving any climate goals.
To put it another way: generative AI, and the construction of the hyperscale data centers that run it, is the latest and most lucrative rationalization for exponential growth in Silicon Valley (and resembles earlier tech bubbles in this and other respects). Tech companies are currently inserting this technology wherever they can think to put it, climate and ethical considerations be damned.
We, as a field, must discuss if and how we accept—and refuse—these tools in our working lives, recognizing and acknowledging their costs and limitations. In this talk, to aid that discussion, I will briefly outline what is known about the carbon emissions associated with generative AI tools and enumerate other ethical considerations about their use, including the provenance of their training data, the deliberately obfuscated human labor required to make their output usable, and their pollution of the information ecosystem.
| Talk category | Sustainability (contact f.f.s.van.der.tak@rug.nl) |
|---|---|
| Preference for a talk or poster | Talk |
