Ekiya Symbolic Reasoning

Symbolic AI is another type of AI, different from neural networks.

Rather than imitating the architecture of the brain, symbolic AI is an approach to mimic logical and mathematical reasoning. For example: logical deduction, theorem proving, constraint propagation or automatic case classification.

One strength of symbolic AI is the traceability and explainability of its results. It is a technology that leaves no room to hallucinations because it is based on logical and formal models.

Our own research has shown that combining different types of AI is often the most efficient approach to provide domain specific AI solutions. One might use symbolic AI to create prompts for a generative AI agent in such a way that leverages on the agent's textual creativity and reasoning power, while limiting the possible range of hallucinations. It is also possible to post-process the agent's answers in order to challenge it against a validated truth model.

There is no free lunch and no AI solution fits all problems even though Gen-AI goes toward a kind of General type of AI. Thus, our approach is to combine agents that are all able to solve very specific problems using their own rules and skills.

These multi-agents systems can generate Graph-Of-Thoughts systems that can address many kinds of problems.