Addressing the Limitations of AI/ML in creating Cognitive Solutions

Abstract In order to solve real-world problems associated with development of Cognitive Solutions, several key problems need to be addressed. The current state of art in AI/ML is too limited for it to offer practical solutions that will outperform manually developed solutions in a large number of contexts. Building upon experience developed in creating AI/ML solutions for enterprise networks and IoT focused solutions, we will outline the limitations of current AI/ML systems, and discuss ways to address those challenges. We would also describe the vision of an AI/ML based system that ought to be developed in order to attain the vision of a true cognitive AI/ML systems that can be used in a broad set of enterprise contexts.
  • Dinesh Verma (IBM US)
  • Archit Verma (IBM US)
  • Utpal Mangla (IBM US)
Date Dec-2021
Venue 2021 IEEE Third International Conference on Cognitive Machine Intelligence (CogMI), pp. 189-196. IEEE, 2021