Collective Superintelligence : Recursion, Iteration, AI Join Groups to Improve NLP Interpretations.

EasyChair Preprint 3380

Abstract

If research produces sufficiently intelligent software, it would be able to reprogram and improve itself, and could continue doing so leading to a superintelligence. A superintelligence may be an emulated human with a human-like reasoner requiring long strings of actions. This raises the possibility of collective superintelligence : a large number of separate reasoning systems could act in aggregate with far greater capabilities than any sub-agent. In this paper, we have implemented collective superintelligence which is an ideal future in which Recursion & Iteration models work along side AI for combining the intelligence of different models along with AI to better NLP interpretations for determining the most likely parse tree for a sentence. When we used other methods along with AI which allowed multiple models to work together for achieving improved interpretations beating those of individual models alone. The results from our study shows superior performance of a combined machine and AI augmented model compared to either AI or machine alone. The combined approach shows the feasibility of our method and could harness the best of machine intelligence & artificial intelligence to create a collective superintelligence.

Keyphrases: Collective Superintelligence, Combined Yield, Recursive Neural Networks, parsing

@booklet{EasyChair:3380,
  author    = {Poondru Prithvinath Reddy},
  title     = {Collective Superintelligence : Recursion, Iteration, AI Join Groups to Improve NLP Interpretations.},
  howpublished = {EasyChair Preprint 3380},
  year      = {EasyChair, 2020}}