cross-posted from: https://hexbear.net/post/9063453
X: @deanwball
Kimi K3, the new model released from a Chinese startup company is so good that it surpasses frontier US Ai models on several benchmarks.
It is nearly a Fable level Ai.
For context Fable was recently restricted by the Trump administration due to national security concerns, that’s how good it is.
The best part is, the Kimi team is releasing the model to the public so that anyone can run it independently. This completely undermines the business model of Anthropic and Openai, and rocked the US stock market.
If it was only local open source models there’s a lot of criticisms that I would keeps my lips shut about just due the provided context that it brings.
Like yeah LLMs are not a legitimate buisness, even with the most basic argument that everyone metions with theft. You can’t have a legitimate buisness entirely based off of stealing literally ALL content and intellectual property rights. You certainly can’t then reatrict access to said information you stole, then start funding lobbies who are pushing for digital ID’s ban of VPNs and what is essentially the DRM of Ai.
And the theft isnt even 1/10th of my biggest gripes with LLMs and these ai companies. Fucking grok will tell you there is no genocide in Israel but there’s one on south African happening to white farmers. They’ll use Ai to ban key words like complaining about Israels control over the us, but they won’t use the Ai to stop csam ads from showing up, no that hurts their revenue they need to pay off the token bill. Oh and if you dislike any of those facts you are a terrorist. Flock has tagged you such and now when ever your car shows up on their cameras or a cop runs your license plate it designates you as one.
I don’t get how he can’t see himself as infaluatable, like his ideals aren’t the evil diatopian future. I’m guessing its because on this scenario he has to live like the surfs because he didnt get his bag before it was too late.
