Conversation

Notices

1. Embed this notice
   FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:03:07 JST FediTips has moved!

   Some apps on here are offering AI/LLM for certain features such as entering an image description automatically.

   If you are abled and sighted, please don't use these features.

   If you are sighted and able to enter descriptions yourself, it would be better for the planet if you enter descriptions manually. Manual descriptions avoid anyone having to cause environmental damage.

   So-called "AI" features damage the planet:

   https://www.technologyreview.com/2023/12/05/1084417/ais-carbon-footprint-is-bigger-than-you-think/

   https://www.nature.com/articles/s42256-020-0219-9

   https://oecd.ai/en/wonk/how-much-water-does-ai-consume

   • Embed this notice
     runarcn (runarcn@mstdn.social)'s status on Thursday, 14-Dec-2023 04:03:48 JST runarcn

     in reply to

     @feditips Could you add an edit to clarify whether or not this applies to the mastodon detect text from image feature?

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:08:53 JST FediTips has moved!

     in reply to

     • runarcn

     @runarcn

     I don't know if that feature works any more? However, I'd imagine it causes extra energy costs of some kind.

     By the way, just to repeat, I'm not saying any of this to people who depend on these features for accessibility. I'm only saying this to people who are able to enter a description manually.

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:16:31 JST FediTips has moved!

     in reply to

     • spaduf

     @spaduf

     You've linked to an article from April 2022.

     Here's an article from December 2023:

     https://www.technologyreview.com/2023/12/05/1084417/ais-carbon-footprint-is-bigger-than-you-think/

   • Embed this notice
     spaduf (spaduf@hachyderm.io)'s status on Thursday, 14-Dec-2023 04:16:32 JST spaduf

     in reply to

     @feditips Heads up, a lot of the work suggesting these outlandish water usage stats have since been debunked as overblown. Here's an article (https://arxiv.org/abs/2204.05149) that discusses some issues in methodology and goes further to point out that a significant portion of the ecosystem still has hardly been optimized. I understand criticisms about inhuman accessibility mechanisms and the potential for error/alienation, but I'm not sure we should be discouraging it on the basis of climate.

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:20:43 JST FediTips has moved!

     in reply to

     • spaduf

     @spaduf

     The article I linked to is specifically about the impact of using the models (not training but actual use):

     "And while we know training AI models is highly polluting, the emissions attributable to using AI have been a missing piece so far. That is, until now. "

     https://www.technologyreview.com/2023/12/05/1084417/ais-carbon-footprint-is-bigger-than-you-think/

   • Embed this notice
     spaduf (spaduf@hachyderm.io)'s status on Thursday, 14-Dec-2023 04:20:44 JST spaduf

     in reply to

     @feditips One HUGE caveat to this is to note the difference between inference and training. The differences in resource usage are fairly massive, but training costs should be thought of as closer to an initial upfront cost rather than an ongoing expense of resources.

   • Embed this notice
     LAUREN (noondlyt@mastodon.social)'s status on Thursday, 14-Dec-2023 04:21:16 JST LAUREN

     in reply to

     @feditips This is akin to crypto's impact on the planet.

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:21:16 JST FediTips has moved!

     in reply to

     • LAUREN

     @noondlyt

     Exactly. Best to avoid it if you can.

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:22:23 JST FediTips has moved!

     in reply to

     • Ken Kinder :clubtwit:

     @bouncing

     If you actually read the article I linked to, you would see it is about energy of actual usage (not training):

     "And while we know training AI models is highly polluting, the emissions attributable to using AI have been a missing piece so far. That is, until now. "

     https://www.technologyreview.com/2023/12/05/1084417/ais-carbon-footprint-is-bigger-than-you-think/

   • Embed this notice
     Ken Kinder :clubtwit: (bouncing@twit.social)'s status on Thursday, 14-Dec-2023 04:22:24 JST Ken Kinder :clubtwit:

     in reply to

     @feditips building the AI model is what takes a lot of resources. Running the model is pretty minimal and often can be done totally locally on something as low-powered as a cell phone.

     I’d say this is an incorrect take. 🤷♂️

   • Embed this notice
     FediTips has moved! (feditips@mstdn.social)'s status on Thursday, 14-Dec-2023 04:37:25 JST FediTips has moved!

     in reply to

     • Ken Kinder :clubtwit:

     @bouncing

     As far as I can tell, Voiceover is just looking for very specific things like text and icons?

   • Embed this notice
     Ken Kinder :clubtwit: (bouncing@twit.social)'s status on Thursday, 14-Dec-2023 04:37:27 JST Ken Kinder :clubtwit:

     in reply to

     @feditips That’s from scratch text-to-image generation, not captioning.

     Eg, VoiceOver can run on iOS and basically describes what you see in your camera app in real time. Were it using as much energy as it takes to charge your phone, I imagine you’d notice.

     I don’t necessarily know what tools what software is using to caption images, but most people can run some form of captioning locally without the CPU breaking a sweat.

   • Embed this notice
     Didier (on8sd@mstdn.social)'s status on Friday, 15-Dec-2023 00:56:45 JST Didier

     in reply to

     @feditips Just thinking: how much CO2 do we produce for publishing 1 toot? For real.