I firmly believe that GPT is just a stepping stone. I think the future for such tools is some form of symbios with humans where systems can be developed that will take full advantage of the goof attributes of each group member.
For example, tools like GPT are great at managing huge amounts of data and their pattern matching capabilities are very good at recognizing relationships that exist in this data, provided therre is significant similiarity between existing data and new data that is newly being analyzed. They are less good at creative imagination of possibilities that have not been seen before.
Humans, on te other hand, excel at this type of creative thinking but have severe limitations on the amount of data that they can handle.
I am not convinced that natural language is always the best medium of communication between humans and computers. I think that we need to devise new ways of communicating that are closer to the problem being examined. In particular, natural language has limitations in communicating complex structural information and our graphical interfaces are currently very human-centric and rely heavily on the byproducts of the perception of vision.
Working to develop these possibilities should provide an interesting time for the near future.
I firmly believe that GPT is just a stepping stone. I think the future for such tools is some form of symbios with humans where systems can be developed that will take full advantage of the goof attributes of each group member.
For example, tools like GPT are great at managing huge amounts of data and their pattern matching capabilities are very good at recognizing relationships that exist in this data, provided therre is significant similiarity between existing data and new data that is newly being analyzed. They are less good at creative imagination of possibilities that have not been seen before.
Humans, on te other hand, excel at this type of creative thinking but have severe limitations on the amount of data that they can handle.
I am not convinced that natural language is always the best medium of communication between humans and computers. I think that we need to devise new ways of communicating that are closer to the problem being examined. In particular, natural language has limitations in communicating complex structural information and our graphical interfaces are currently very human-centric and rely heavily on the byproducts of the perception of vision.
Working to develop these possibilities should provide an interesting time for the near future.