02-06-2025, 06:14 AM
I don't doubt that these systems do work in very specific tasks but when they get to a more generalised nature I have found that results can very easily go haywire if your input is not in the norm of their acceptance criteria.
I don't know how much is due to actual AI and how much to advanced search (although AI could be expanded to include human aided AI) but I find that finding information these days can be a pain as search engines gear themselves to intelligent responses to queries in a general way but this fact makes it difficult for me to find information of a specific nature which I know exists but is not mainstream. As in the example here, the response appears to be generally simplistic rather than attempting to give what is really needed. I feel it often makes getting the result needed much harder than it would be without. There maybe needs to be different intelligence levels of AI for different academic levels?
My thoughts in general are to beware of the idea that these systems don't tire or go wrong too. Software is often still bug ridden and the more complex it is the greater the chance of a glitch. Don't forget that your Windows computer will generally slow down as time progresses, for example. The hardware can fail in strange ways too. An AI system is a complex system as is a human. The more complex it becomes the more un-predictable it will be. Dedicated AI is easier to predict but people will attempt to enslave all forms to their aims.
I was involved a bit in the early days of AI and do remember some amusing quirks when the AI responded to things it shouldn't have even tried to understand but was programmed for a broad range of input.
I don't know how much is due to actual AI and how much to advanced search (although AI could be expanded to include human aided AI) but I find that finding information these days can be a pain as search engines gear themselves to intelligent responses to queries in a general way but this fact makes it difficult for me to find information of a specific nature which I know exists but is not mainstream. As in the example here, the response appears to be generally simplistic rather than attempting to give what is really needed. I feel it often makes getting the result needed much harder than it would be without. There maybe needs to be different intelligence levels of AI for different academic levels?
My thoughts in general are to beware of the idea that these systems don't tire or go wrong too. Software is often still bug ridden and the more complex it is the greater the chance of a glitch. Don't forget that your Windows computer will generally slow down as time progresses, for example. The hardware can fail in strange ways too. An AI system is a complex system as is a human. The more complex it becomes the more un-predictable it will be. Dedicated AI is easier to predict but people will attempt to enslave all forms to their aims.
I was involved a bit in the early days of AI and do remember some amusing quirks when the AI responded to things it shouldn't have even tried to understand but was programmed for a broad range of input.
