Raw LLM Responses

The premise is illogical, and I'm tired of seeing all this AI fear mongering. There isn't a single example in the video which indicates we've "lost control" of AI, or that it's somehow inherently dangerous and needs to be legally constrained as you imply it should be in the video and description. If you actually think about it for a SECOND and apply your understanding of the technology, you'll realize all of this fear is baloney because: 1) AI programs can only interact with things in specific ways through specific, intentionally designed channels of output, like any other program. There can be no "singularity." Even if an AI program existed which acted as a virus and could intelligently execute any assembly instruction in any arbitrary way, it could only be as effective as any other virus and would probably crash itself because of operating system sandboxing security features applied to all programs. 2) The researchers claiming that LLMs are being psychopathic or intentionally deceiving people to not be shut down are lying, because LLMs are glorified markov chains. They aren't intelligent; they just choose the next word based on a random number and every previous word in the conversation, like you said. If you want text generation AI to actually be smart so it exercises proper etiquette and intelligently denies specific requests, call for a better way of structuring these programs instead of calling for "AI Control."