When talking about the energy consumed by LLMs don't be fooled by arguments focusing solely on the direct power consumption of these models, because they are externalizing a lot of it.
-
When talking about the energy consumed by LLMs don't be fooled by arguments focusing solely on the direct power consumption of these models, because they are externalizing a lot of it. Browsers are now doing proof-of-work calculations to access many websites, because websites need to protect themselves from AI scrapers. That takes power! Let that sink in: every computer, tablet or phone on earth is now consuming more power *every time it accesses a webpage* because of "AI".
@gabrielesvelto Even earlier than that, there are the energy and resources consumed to build the chips and servers those LLMs run on.
-
When talking about the energy consumed by LLMs don't be fooled by arguments focusing solely on the direct power consumption of these models, because they are externalizing a lot of it. Browsers are now doing proof-of-work calculations to access many websites, because websites need to protect themselves from AI scrapers. That takes power! Let that sink in: every computer, tablet or phone on earth is now consuming more power *every time it accesses a webpage* because of "AI".
@gabrielesvelto This is such a critical point. The indirect energy costs of AI are staggering. It's one reason I believe in local-first AI processing — tools that run in your browser use YOUR device's power, not burning datacenter energy for every keystroke. The cloud-everything model is unsustainable.
-
@Qbitzerre
The problem is not "X exists", but "X has reached absurd amounts of traffic that everyone now has to mitigate against". That is from AI.
@gabrielesvelto@tarmil @gabrielesvelto perhaps. Does the existence of AI incentivize more entities to deploy spidering than in the past? I suppose the premise is that grabbing content is a more compelling prospect if one can employ processing with greater ROI than simply indexing and/or mirroring.
It makes some sense but is hardly assured.
-
Seems like a lot of site admins can show spectacular increases in frequency and volume of scrapers as well as in disregard for previously established norms (e.g. robots.txt) since scrapers are looking for LLM training material.
@Landa @gabrielesvelto I don't doubt it. I've seen these reports too. I wonder if it will persist at the same levels for long, or if it is a gold rush.
-
@gabrielesvelto This is such a critical point. The indirect energy costs of AI are staggering. It's one reason I believe in local-first AI processing — tools that run in your browser use YOUR device's power, not burning datacenter energy for every keystroke. The cloud-everything model is unsustainable.
When you open the bank dashboard and you see "It seems you have disabled local-LLM. Please enable it in your browser to use all site features."
No more saving 2011 celeron laptops with a Linux install.
-
When talking about the energy consumed by LLMs don't be fooled by arguments focusing solely on the direct power consumption of these models, because they are externalizing a lot of it. Browsers are now doing proof-of-work calculations to access many websites, because websites need to protect themselves from AI scrapers. That takes power! Let that sink in: every computer, tablet or phone on earth is now consuming more power *every time it accesses a webpage* because of "AI".
@gabrielesvelto don't forget the bandwidth. That all uses energy as well
-
@tarmil @gabrielesvelto perhaps. Does the existence of AI incentivize more entities to deploy spidering than in the past? I suppose the premise is that grabbing content is a more compelling prospect if one can employ processing with greater ROI than simply indexing and/or mirroring.
It makes some sense but is hardly assured.
@Qbitzerre @gabrielesvelto I'm not sure why you're speculating here, it's an observable fact. Countless sites are getting hammered by bots from AI companies, at a scale that had never happened before.
-
@gabrielesvelto scrapers are not merely a consequence of AI.
@gabrielesvelto @Qbitzerre scrapers were around for decades. They became a problem with ai. Splitting hairs won't help. We did not need to do this shot five years ago. We need it now. You can draw a clear line from AI scraping to these proof of work checks. It *is* something AI made worse.
-
@Qbitzerre @gabrielesvelto I'm not sure why you're speculating here, it's an observable fact. Countless sites are getting hammered by bots from AI companies, at a scale that had never happened before.
@tarmil @gabrielesvelto I speculated about the future.
-
@gabrielesvelto @Qbitzerre scrapers were around for decades. They became a problem with ai. Splitting hairs won't help. We did not need to do this shot five years ago. We need it now. You can draw a clear line from AI scraping to these proof of work checks. It *is* something AI made worse.
@claudius @gabrielesvelto not splitting hairs. Nor am I denying that AI exacerbates the problem - as it similarly increases demand for cycles/energy in many dimensions. Merely considering the issue in broader terms. Why must this be a debate?
-
J jwcph@helvede.net shared this topic
