LLM’s don’t operate in a vacuum anymore. They’re given tools and access to systems.
An LLM could for sure shut down a database if you ask it to reduce costs and you give it database management tools. That’s getting closer to the paperclip situation.
…still not how this works. The model itself would need reasoning capability. The ability to regurgitate the story does not make it capable of taking the necessary actions.
A computer automated system doesn’t currently exist, obviously. That would be reckless and pointless, so thankfully, nobody’s seen a reason to implement it.
I just know techbros and similar will want to build and connect LLMs to systems like that as soon as they think they can get away with it.
The paperclip scenario requires an AI capable of reasoning - LLMs are nowhere near that.
LLM’s don’t operate in a vacuum anymore. They’re given tools and access to systems.
An LLM could for sure shut down a database if you ask it to reduce costs and you give it database management tools. That’s getting closer to the paperclip situation.
If you give it access it’s at least partly on you.
Unfortunately, we’ve written a lot about the paperclip scenario, I wouldn’t be surprised if it was in the training data
…still not how this works. The model itself would need reasoning capability. The ability to regurgitate the story does not make it capable of taking the necessary actions.
Not if we connect it to a bunch of systems that take its regurgitations as instructions
Find me an example of a system that can translate “make this steel mill produce paperclips” into actions. I’ll wait.
A computer automated system doesn’t currently exist, obviously. That would be reckless and pointless, so thankfully, nobody’s seen a reason to implement it.
I just know techbros and similar will want to build and connect LLMs to systems like that as soon as they think they can get away with it.