A brief introduction to Symbolic AI: What? Why? How?
Understanding the basics of Symbolic AI
Currently, artificial intelligence is dominated by neural networks, machine learning, and deep learning. But earlier, it was all about classical AI or Good, Old-Fashioned AI (GOFAI), or popularly known as symbolic AI. This branch of artificial intelligence concerns itself with attempting to explicitly represent human knowledge in a declarative form (i.e., facts and rules). This concept differs from machine learning as it involves human intervention, unlike the latter that relies on self-learning rules as it establishes correlations between inputs and outputs. To build a symbolic reasoning system, humans must first learn the rules by which two phenomena relate and then hard-code those relationships into a static program.
How does it work?
Symbolic AI processes strings of characters that represent real-world entities or concepts. These strings are then stored manually or incrementally in a Knowledge Base (any appropriate data structure). They are made available to the interfacing human being/machine as and when requested and used to make intelligent conclusions and decisions based on the memorized facts and rules put together by propositional logic or first-order predicate calculus techniques. This form of AI requires no training, no massive amounts of data, and no guesswork. The best part about it is that unlike the black box of machine learning, it can explain its decisions by showing which parts were evaluated as true or false.
Why it isn’t Popular
This technology is convenient for settings where the rules are simple and straightforward, and users can easily obtain input and transform it into symbols. Hence, Symbolic AI had limited success and, by and large, has left the field to neural network architectures that mimic human intelligence. Another issue encountered in symbolic reasoning is that the computer itself doesn’t know what the symbols mean. This means they are not necessarily linked to any other representations of the world, unlike neural nets, which can link symbols to vectorized representations of the data, which are, in turn, just translations of raw sensory data. This makes it bulky and difficult to set up, for it requires facts and rules to be explicitly translated into strings and then provided to a system.
Enter Neuro Symbolic AI
The aforementioned facts do not mean symbolic AI is dead. In fact, modern and successful applications are based on combined techniques from Symbolic AI, Neural Networks, and Machine Learning. Also, several efforts are being made to combine neural networks and symbolic AI. One such project is the Neuro-Symbolic Concept Learner (NSCL), a hybrid AI system developed by the MIT-IBM Watson AI Lab. A neuro-symbolic system applies logic and language processing to answer a question similar to how a human would reason. NSCL uses both rule-based programs and neural networks to solve visual question-answering problems. Unlike pure neural network-based models, hybrid AI can learn new tasks with fewer data and is explainable. And unlike symbolic-only models, NSCL doesn’t struggle to analyze the content of images.