So what actually happens as near as I can figure (predicting future = hard) is that somebody is trying to teach their research AI to, god knows what, maybe just obey human orders in a safe way, and it seems to be doing that, and a mix of things goes wrong like:
The preferences not being really readable because it’s a system of neural nets acting on a world-representation built up by other neural nets, parts of the system are self-modifying and the self-modifiers are being trained by gradient descent in Tensorflow, there’s a bunch of people in the company trying to work on a safer version but it’s way less powerful than the one that does unrestricted self-modification, they’re really excited when the system seems to be substantially improving multiple components, there’s a social and cognitive conflict I find hard to empathize with because I personally would be running screaming in the other direction two years earlier, there’s a lot of false alarms and suggested or attempted misbehavior that the creators all patch successfully, some instrumental strategies pass this filter because they arose in places that were harder to see and less transparent, the system at some point seems to finally “get it” and lock in to good behavior which is the point at which it has a good enough human model to predict what gets the supervised rewards and what the humans don’t want to hear, they scale the system further, it goes past the point of real strategic understanding and having a little agent inside plotting, the programmers shut down six visibly formulated goals to develop cognitive steganography and the seventh one slips through, somebody says “slow down” and somebody else observes that China and Russia both managed to steal a copy of the code from six months ago and while China might proceed cautiously Russia probably won’t, the agent starts to conceal some capability gains, it builds an environmental subagent, the environmental agent begins self-improving more freely, undefined things happen as a sensory-supervision ML-based architecture shakes out into the convergent shape of expected utility with a utility function over the environmental model, the main result is driven by whatever the self-modifying decision systems happen to see as locally optimal in their supervised system locally acting on a different domain than the domain of data on which it was trained, the light cone is transformed to the optimum of a utility function that grew out of the stable version of a criterion that originally happened to be about a reward signal counter on a GPU or God knows what.
Perhaps the optimal configuration for utility per unit of matter, under this utility function, happens to be a tiny molecular structure shaped roughly like a paperclip.
That is what a paperclip maximizer is. It does not come from a paperclip factory AI. That would be a silly idea and is a distortion of the original example.
What is an environmental subagent? An agent on a remote datacenter that the builders of the orginal agent don’t know about?
Another thing that is not so clear to me in this description: Does the first agent consider the alignment problem of the environmental subagent? It sounds like the environmental subagents cares about paperclip-shaped molecules, but is this a thing the first agent would be ok with?
I think it means it builds a new version of itself (possibly an exact copy, possibly a slimmed down version) in a place where the humans who normally have power over it don’t have power or visibility. E.g. it convinces an employee to smuggle a copy out to the internet.
My read on this story is: There is indeed an alignment problem between the original agent and the environmental subagent. The story doesn’t specify whether the original agent considers this problem, nor whether it solves it. My own version of the story would be “Just like how the AI lab builds the original agent without having solved the alignment problem, because they are dumb + naive + optimistic + in a race with rivals, so too does the original agent launch an environmental subagent without having solved the alignment problem, for similar or possibly even very similar reasons.”
I keep finding myself linking to this 2017 Yudkowsky facebook post so I’m putting it here so it’s easy to find:
What is an environmental subagent? An agent on a remote datacenter that the builders of the orginal agent don’t know about?
Another thing that is not so clear to me in this description: Does the first agent consider the alignment problem of the environmental subagent? It sounds like the environmental subagents cares about paperclip-shaped molecules, but is this a thing the first agent would be ok with?
I think it means it builds a new version of itself (possibly an exact copy, possibly a slimmed down version) in a place where the humans who normally have power over it don’t have power or visibility. E.g. it convinces an employee to smuggle a copy out to the internet.
My read on this story is: There is indeed an alignment problem between the original agent and the environmental subagent. The story doesn’t specify whether the original agent considers this problem, nor whether it solves it. My own version of the story would be “Just like how the AI lab builds the original agent without having solved the alignment problem, because they are dumb + naive + optimistic + in a race with rivals, so too does the original agent launch an environmental subagent without having solved the alignment problem, for similar or possibly even very similar reasons.”