No, LLMs are not sentient and never will be. The narratives that the AI companies are pushing out are marketing stunts designed to get people talking about false concerns and provide an argument for them to lower their training costs since they are running out of capital to burn.

in the broadest of strokes, no. Consumer level AI is more or less a toddler who has had far far too much information dumped into it. At best it's a competent "Yes man", and worst, it's actively detrimental to whatever people are attempting to do with it.

As far as I can reason, in order for an AI to get to those levels of intelligence, it needs a few things it currently lacks entirely.

• Proper information compartmentalization and connecting strings to get from one to another in contextually relevant ways

• A small cabal of other AIs to determine both the intent of the prompt it's being given, and that the response is appropriate and/or factual where needed.

• Some level of curiosity to seek out answers to prompts and use compartmentalizes those answers for future recall.

If an AI had these abilities, it would approach becoming a proper General AI that if given the proper resources could potentially achieve singularity.

No. The closest thing is essentially a hallucinating facial recognition bot with a mounted gun. Dangerous but ultimately directionless, as facial recognition technology still makes huge mistakes.

You just haven't given humans enough time. Look these idiots are working on it now:

https://www.tomshardware.com/tech-industry/drones/autonomous-strike-drone-uses-nvidia-jetson-orin-nano-to-independently-pick-and-bomb-targets-swedish-startups-attack-drones-run-small-ai-model-require-no-human-input-and-zero-external-comms

There would have been, but a warrior called Kyle Reese traveled back in time from the future, and informed Sarah Connor of the coming future, and she took that shit down

Is current AI sentient as the Terminator?

No, it's not as capable by a long shot.

It's hard to try to come up with a single measurement of capability---I suspect that if we had a really good one, we'd know more than we do about intelligence. But one can't reasonably call the current crop general intelligences. You can't just feed them arbitrary problems and have them learn to solve them and get better at doing so over time, the way humans can. They have quite limited ability to learn subsequent to their training phase.

EDIT: My own take is that while an artificial general intelligence might incorporate an LLM in some way, it won't be just an LLM. Like, my guess would be that we need foundational changes to how the things work to get there.

LLMs can do some neat things, and I'm sure that people will figure out more clever applications. But I don't believe that it'll be what we have today or a minor tweak to it or throwing more hardware at it that'll get us to general intelligence. Going to take more time with researchers advancing the field.

The software is there. The problem is hardware. The best in the world is stuff like the Boston Dynamics quadrupeds. They're good but not great. We can't quite replicate fast, fine motor function, especially for grasp-but-not-crush hands. And stuff like that burns battery power far too quickly for what modern battery packs can provide.

What's worse-- the very real threats of AI, or nonsense / BS posts like this, written by complete idiots?