DailyPost 2266
ARTIFICIAL GENERAL INTELLIGENCE – HOW FAR OR HOW CLOSE?
The release of ChatGPT on 30th Nov. 2022 and its massive following of over 1 million in just a short time, once again raises the question as to how are far are we from artificial general intelligence (AGI). It has bene reasonably accepted that it cannot happen as a part of one research project. It could happen with lots of projects in different companies /sectors /geographies adding up into the even now seemingly elusive AGI. AGI means the representation of generalized human cognitive abilities is software. What is actually connotes it that faced with an unfamiliar task, the AGI system would be able to find a solution.
How much would the world would take to reach AGI has been rattling researchers alike for quite a long time. It has never gained too much attention because it was believed to beyond a decade away. It might not be a clear-cut indication of any timeline of the AGI in the making, but the release and super success of ChatGPT, we can safely presume that the timeline has been shortened considerably. Users have found the performance of ChatGPT based on GPT 3.5 and being a sibling of InstructGPT ranging from awesome to scary. The sum and substance of it is that people are finding a tool, unlike any other from the AI basket on offer.
Is ChatGPT human like or beyond humanlike, the way humans are responding to it gives us a feeling that what they are getting is more than what they expect out of above average human being. Content creation for a variety of tasks, based on discussion and dialogue is something we did not expect from a machine in our lifetime. To top it all, is codes in a manner that it throw off most of the coders out of job. Even best in the field found the instantaneous churned out codes mindboggling. Coding will never be the same again is the promise of ChatGPT. It is the best large language dialogue based all-purpose model. Very rare you find it off the mark on content and coding compared to what an average person would churn out.
These are very early days, if it could learn so much is so to say lab conditions, what would be it learning with millions of users using it day in and day out. It would be an exponential leap, which we find difficult to envisage as it stands today. We have a limitation as nothing of this nature as ever been experienced before. For the model as well, it is a great real-life learning. The scale of scaling no one can predict. It seems in all likelihood in the areas which this tool is in, AGI is not a far way off. How we define is for the machine is finally left to us. There are lots of AI tools and its applications that breaking barriers elsewhere too. How soon and how much will these all converge is the question? Herein lies the timeline of the final arrival of AGI, ushering in new world for humankind.
AGI FROM A DISTANT FUTURE SEEMS TO COME ON THE HORIZON OF A FORSEEABLE FUTURE.
Sanjay Sahay