#3 | Posted by oneironaut
I think you might want a redo on that first full sentence.
I agree we need to build AIs for the future. The fact is things like Kimi will appear. I don't know much about Kimi but I am usually skeptical of anything Chinese. (Remember for example that OpenAI says DeepSeek 'inappropriately' copied ChatGPT...)
Also on Kimi K3 from what I have been seeing, they are just comparing it to Fable and on coding. It sounds as if it is comparable to Sonnet 5 price wise but Sonnet is faster... Anthropic keeps tuning their older engines like Sonnet and Haiku.
As for this is funded by China. It is possible but folks like you don't want to hold foreign countries tampering in the US accountable. When you say "environmental groups" - that is not in Russia's interest to fund. Perhaps China's. Russia prefers to fund right leaning propaganda and the NRA. These Mega data centers are a HUGE problem. AI isn't in the hands of everyone in China - keep that in mind.
Another example of emerging things was done by a Netflix engineer - Tejas Chopra. He created an open-source tool called Headroom that compresses AI agent inputs (such as logs, tool outputs, code, and conversation history) before they reach the language model. This context compression layer can reduce token usage by 60% to 95% while maintaining output quality.
Just like everything else where there is, well, literally anything there is abuse.