3 ms·
The ad model is no longer needed, or even desired. When the operational expense is so low due to a simple LLM and very few employees, other business models are
by ldjkfkdsjnv 3y ago
The ad model is no longer needed, or even desired. When the operational expense is so low due to a simple LLM and very few employees, other business models are possible.
- endisneigh 3y agoNeeded or desired by whom? It’s easy to make money with ads, therefore it will happen. Energy isn’t free and people won’t pay. The hypothetical AGI will necessarily require a lot of compute so only big companies will have access. Funny to see people rooting for meta. Did you forget how they make money?
- vdaea 3y agoTraining LLM models and generating LLM responses is probably more expensive than... basically anything else we have now.
- ldjkfkdsjnv 3y agoIts expensive right now. Compute cost always drops like a rock
- refulgentis 3y agoIt's quietly broken in favor of local LLMs over the past month. I say this as someone who was a huge skeptic until maybe 2 weeks ago. It's not GPT-4 but it doesn't need to be. StableLM 3B can handle RAG inputs. This is huge. 7B models can't run on consumer mobile hardware: https://x.com/jpohhhh/status/1747451790969184682?s=20 https://x.com/jpohhhh/status/1747451790969184682?s=20 llama.cpp got examples for iOS / Android in December. It can run on all platforms via one library: https://x.com/jpohhhh/status/1748852554920849579?s=20 https://x.com/jpohhhh/status/1748852554920849579?s=20 retrieval/vector DB can run on all platforms via one library: https://github.com/Telosnex/fonnx https://github.com/Telosnex/fonnx
- smoldesu 3y ago> The ad model is no longer needed, or even desired. I'll believe that when YouTube is dead and gone.