![[-]](https://www.aiowares.com/images/collapse.png "[-]")
03-11-2023, 11:13 PM
I tried ChatGPT and to be honest I found the answers to my general questions quite good. I must admit it's a remarkable IT achievement. However, what people don't realize it is like a long-running stand up comedy show. If you are watching for the first time you may be very surprised and think that the comedian is a true god of improvisation when actually each joke had to gradually polished until it reached its perfection.
That is the AI training. Researchers, engineers, analysts spend years training AI with some "photograph" of the knowledge taken at some point in time. When you see it for the first time you think it is a true marvel and suddenly you start projecting all the potential that it could reach, ignoring that it could take years to train (have significant amount of training data organized and classified by humans for instance) it to do all the stuff you are expecting.
This is not a thing restricted to chatbots nor AI. In fact, every product announcement is very well rehearsed to provoke strong reaction in consumers. They are not always lying us about their product capabilities. They only need to omit few facts and let our imagination do the job.
That is the AI training. Researchers, engineers, analysts spend years training AI with some "photograph" of the knowledge taken at some point in time. When you see it for the first time you think it is a true marvel and suddenly you start projecting all the potential that it could reach, ignoring that it could take years to train (have significant amount of training data organized and classified by humans for instance) it to do all the stuff you are expecting.
This is not a thing restricted to chatbots nor AI. In fact, every product announcement is very well rehearsed to provoke strong reaction in consumers. They are not always lying us about their product capabilities. They only need to omit few facts and let our imagination do the job.
