You are right I need to differentiate terms.
It's not all "AI". But it heavily utilize AI-type (massively parallel tensor distillation) technologies, and the hype surrounding these technologies.
Intel/Microsoft
- They normalize layered lithography, that transistors cannot be visually verified to perform the advertised function.
- They implement "NPU" in new processors. Ultra.
- They backdoor Copilot into email.
- They use OCR on screenshots of your workspace and share to cloud with Recall.
Primarily this would be the colloquial "AI" danger.
Some of what I meant is more like deepfaked presidential addresses and business phone calls. Entire news networks that are not humans. Perpetually training on real-time data, decide and execute maneuvers that lead toward ::goal::
Some of what I meant is a giant LLM input your entire internet history and now can stable diffuse accurately what your response will be to any situation. To utilize, they query the action they want you to perform and receive the set of articles you'd have to read in order to perform that action.
AND I assert with the "salt" statement before, that you can only recognize LLM speech because it is trained on the post-salted internet. The largest unsalted LLM on earth is indistinguishable from human speech.
And without much evidence I believe it was born in a Bitcoin mining facility.
You are right I need to differentiate terms.
It's not all "AI". But it heavily utilize AI-type (massively parallel tensor distillation) technologies, and the hype surrounding these technologies.
Intel/Microsoft
- They normalize layered lithography, that transistors cannot be visually verified to perform the advertised function.
- They implement "NPU" in new processors. Ultra.
- They backdoor Copilot into email.
- They use OCR on screenshots of your workspace and share to cloud with Recall.
Primarily this would be the colloquial "AI" danger.
Some of what I meant is more like deepfaked presidential addresses and business phone calls. Entire news networks that are not humans. Perpetually training on real-time data, decide and execute maneuvers that lead toward ::goal::
Some of what I meant is a giant LLM input your entire internet history and now can stable diffuse accurately what your response will be to any situation. To utilize, they query the action they want you to perform and receive the set of articles you'd have to read in order to perform that action.
AND I assert with the "salt" statement before, that you can only recognize LLM speech because it is trained on the post-salted internet. The largest LLM on earth is indistinguishable from human speech.
And without much evidence I believe it was born in a Bitcoin mining facility.