Experts agree nuclear will take anywhere from 10 or more years to scale and right NOW the water being used for LLM tools that,
@GaryMarcus and other trusted experts remind us DAILY are more flawed than helpful overall, is NOT justified. Period. We are WASTING water NOW.
LLMS
-
LLM Water Waste Unjustified Until Nuclear Energy Scales
By
–
-
LLMs Water Consumption and Energy Sustainability Concerns Criticized
By
–
The world NEEDS water. Where non LLM oriented AI is helping the planet (and it is in many ways), yes – let’s use it. Where the ecosystem surrounding LLMs including ye olde chip logic “justified” by saying “nuclear fusion will solve our energy needs” is again obfuscation .
-
LLMs versus cattle industry environmental impact comparison ethics
By
–
Then people say, but the cattle industry is worse than LLMs” which I don’t necessarily disagree with, EXCEPT that analogy is akin to saying, “gun deaths by X people is still more than Y people” while inferring or overtly supporting the second scenario!!
-
Beyond LLMs: Alternative Technologies Driving AI Progress Forward
By
–
Similarly, it seems that even if LLMs hit a major technological roadblock (model collapse, inherent limits to the technology) that will not bring development to a halt. Many other technologies are waiting in the wings, and there are other ways to improve LLM performance as well.
-
AI Water Usage Transparency Gap Raises Responsibility Questions
By
–
I’m not trying to condemn DC owners re their lack of reporting, but want to make a point related to your excellent reporting: Nobody can say ANY AI is “responsible” or “for good” where so much water is being used for AI/LLMs while so little data is available on actual use.
-
AI Model Selection: The New Coding Skill
By
–
Yea. Maybe the real skill that will replace coders will be knowing which AI model is best for every potential scenario.
-
Matryoshka Retriever: High Performance Vector Search Technique
By
–
Matryoshka Retriever A recent blog post by @supabase described a new technique for higher performance retrieval, without compromising accuracy. So we've just released a new retriever in LangChain.js which implements this exactly! Use any vector store, and two embedding
-
ChatGPT Working on Reversion History for Custom GPTs
By
–
WIP: ChatGPT is working on reversion history for custom GPTs so it will be simpler to roll back
-
ChatGPT phasing out Plugins in favor of GPTs
By
–
WIP: ChatGPT is preparing GPT Plugins for the sunset by suggesting to explore GPTs instead
-
Query Analysis Techniques for RAG Systems with LLMs
By
–
Query Analysis In OpenAI's retrieval talk on DevDay, they mention a bunch of strategies they experimented with 3 of these can be classified as **query analysis** a RAG technique that is becoming popular to do with LLMs We've added a docs deep-dive on this We cover six