Cognitive Automation and the Disruption of Business Services
Legal Knowledge Management and the Rise of Artificial Intelligence
Robotic Refactoring the Workplace
Why Your Next Insurance Claims Processor Could be a Robot
Building an AI Based Machine Learning for Global Economics
Alexander Fleiss, CIO & CEO, Rebellion Research Partners LP
The Forgotten Element in Your Big Data Strategy
HK Bain, CEO, Digitech Systems
Natural Language Processing and the Future of Customer Service
Eugene Willard, CIO, LaserShip, Inc.
AI is not just for Silicon Valley: Powering Changes in the Energy...
Chris Shelton, President, AES Next, The AES Corporation
Thank you for Subscribing to CIO Applications Weekly Brief
Common Sense Reasoning in AI
Artificial Intelligence, deep learning, and machine learning are interrelated and built on different layers of abstractions. AI is the development of machines mimicking human intelligence to perform tasks. Speech recognition, language translation, visual perception are all examples of AI. Both machine learning and deep learning are subspaces of AI. Machine learning uses artificial neuro-networks and deep learning uses deep neural networks to learn patterns from massive amounts of data.
Deep learning is the ruling factor of AI. In recent years it exploded into the mainstream becoming the dominant way to help machines sense and perceive the world around.
Now top AI researchers are beginning to believe deep learning may not be the correct path to artificial general intelligence (AGI). AGI is the synonym of future general-purpose learning algorithms that can produce human-level cognition. Deep learning relies heavily on large quantities of data and cannot generalize from a few examples, the divergence from the way the human brain learns. It is important to develop an AI with the human ability to model the world. Humans learn and predict using scientific laws that are translated into common sense. Researchers found it difficult to model deep learning with the physical and social model of the world. A deep learning neural network can only label all subjects in a scene but cannot infer the relative relationships between the objects. Inability to adapt non-verbal cues such as body language, facial expressions proves it difficult to build the social model too.
But working together of deep learning and model engines can provide a simplified framework similar to that in the brain. This deep learning provides pattern identification required for labeling objects and models provide the foundation for the logic needed for commonsense reasoning. Researchers are in the initial stages of combining physical and social models to build a cognitive system. Once the labels are stored in a memory location, AI can process them as symbols and then come to a prediction.
The future progress of AI is marked with this integration of social and physical models helping to form a cognitive system, that is common sense.