It’s rather “brave” to make predictions so far out, especially in a rapidly evolving field. The main flaw here I would reckon, from reading Gary Marcus, is the assumption that there’s an easy jump up from the current LLM based AI (souped up predictive text) to Advanced General Intelligence, when it’s more like a quantum leap that there is no guarantee of happening at all.
The interesting part of this is that the prediction was made by an AI. How far AGI is needed for the prediction to be accomplished is a matter for debate, even existing AIs might take us a long way along the path outlined by ChatGPT. Certainly epistemic drift is already happening and MPs are handing analysis over to AIs.
My modification of ChatGPTs's prediction is feasible even with current technology.
It’s rather “brave” to make predictions so far out, especially in a rapidly evolving field. The main flaw here I would reckon, from reading Gary Marcus, is the assumption that there’s an easy jump up from the current LLM based AI (souped up predictive text) to Advanced General Intelligence, when it’s more like a quantum leap that there is no guarantee of happening at all.
The interesting part of this is that the prediction was made by an AI. How far AGI is needed for the prediction to be accomplished is a matter for debate, even existing AIs might take us a long way along the path outlined by ChatGPT. Certainly epistemic drift is already happening and MPs are handing analysis over to AIs.
My modification of ChatGPTs's prediction is feasible even with current technology.
Is A.I. so much scrap iron if someone pulls the plug out?
It is but we are happily pushing the plugs in at the moment. The key to ChatGPT's prediction is that we are the accomplices in our own downfall.
Military AIs operate off grid and dont have plugs, the new anti-personnel AIs have solar charging and wait, hidden, until a target passes.