Email or username:

Password:

Forgot your password?
DW's posts Post Back to profile
DW Innovation

Generative AI not only consumes a lot of energy, it's also likely to produce a lot of electronic trash.

According to researchers, the total amount of global e-waste could grow by the alarming factor of 1.000 in only 5 years.

There are solutions to reducing e-waste – e.g. by prolonging, reusing, and recycling hardware. Their implementation needs global strategies, of course.

An article by DW's Fred Schwaller explains the issue in more detail:

dw.com/en/e-waste-from-ai-comp

#AI #eWaste #environment

11 November at 10:39 | Open on mastodon.social
9 comments
✨🏳️‍⚧️Timelordiroh :she_her:🇵🇸

@dw_innovation Just what we need, yet another reason to hate #GenAI

11 November at 10:42 | Open on mstdn.iroh.tv
Blumine

@dw_innovation

👍 👍

And it isn't the only danger caused by AI:

The images of Spain’s floods weren’t created by AI. The trouble is, people think they were

theguardian.com/commentisfree/

11 November at 10:57 | Open on troet.cafe
Pait

@dw_innovation My 8 year old iPhone says amen. But it's also asking for retirement.

11 November at 11:15 | Open on mastodon.social
mafe en españa

@dw_innovation Let's ask AI how we can cope with all that e-waste!
If it says "ship to Africa like you always do" we know exactly how stupid AI is.

11 November at 11:21 | Open on layer8.space
Nina Kalinina

I no like e-waste, I no like LLM, but the numbers in the DW article do not look right. As per WHO, 62 million tons of waste were created in 2022; so cumulative 5 million tons by 2030 could not mean "around 1,000 times more e-waste than was produced in 2023".

Interestingly, the paper argues we need to "prolong, reuse, and recycle generative AI hardware", and I think we probably should not have manufactured it in the first place.

WHO: who.int/news-room/fact-sheets/

mastodon.social/@dw_innovation

I no like e-waste, I no like LLM, but the numbers in the DW article do not look right. As per WHO, 62 million tons of waste were created in 2022; so cumulative 5 million tons by 2030 could not mean "around 1,000 times more e-waste than was produced in 2023".

Interestingly, the paper argues we need to "prolong, reuse, and recycle generative AI hardware", and I think we probably should not have manufactured it in the first place.

11 November at 11:43 | Open on tech.lgbt
Wolf_Baginski

@dw_innovation
The article feels like a confused mix-up between the AI server centres — factory sites — and the toolbox equivalents suxh as smartphones and tablets. And then there is the e-waste that has been in motor cars since electric starters came into use.

We're not doing enough about any e-waste, and we have too many things built with no attention to recycling. How easy is it to separate a Li-ion battery from a tablet?

And can the expansion percentage from the server farms really be applied to everything?

@dw_innovation
The article feels like a confused mix-up between the AI server centres — factory sites — and the toolbox equivalents suxh as smartphones and tablets. And then there is the e-waste that has been in motor cars since electric starters came into use.

We're not doing enough about any e-waste, and we have too many things built with no attention to recycling. How easy is it to separate a Li-ion battery from a tablet?

11 November at 11:51 | Open on meow.social
Mike Olson

@dw_innovation According to a GenAI pitch deck I saw a couple of months ago, the energy required by GenAI globally doubles every 100 days.

Exponential, on a short X axis interval.

11 November at 14:19 | Open on mastodon.social
Weltzeitgeist

@dw_innovation factor 1000 in 5 years you say? Sheeesh, that's a lot, isn't it?

11 November at 14:33 | Open on norden.social
OnePlanet 💖🇺🇸 🌎

@dw_innovation

What exactly HASN’T “escalated out of control”?

11 November at 14:37 | Open on ohai.social
Go Up