This is from a NYT column earlier today. I make a very astute suggestion just afterwards.
But it’s true, Cade writes, that researchers are transforming chatbots like ChatGPT into systems that can take actions based on the text they generate. A project called AutoGPT can be given a goal like “create a company” or “make some money.” Then it will keep looking for ways of reaching that goal, using other internet services.
Such systems do not work well right now. But in time, those limitations could be fixed.
My comment to the corporate world, there is a much faster, less expensive and far less risky way to make all of this work. Yes just code it yourselves. Do not depend on a system that is going to get all of it wrong for a few years and then some of it possibly right later on…perhaps. And when the dumb system needs updating the AI will in all likelihood start to make larger mistakes again.
But we keep telling you that and now we are seeing just how bad AI is at doing anything at all.
Outside of making a video game run a bit smoother.
The Mechanical Turk