Generative AI has become remarkably good at understanding language, generating content, and solving increasingly complex problems. But there is a fundamental limitation: generating a plausible answer is not the same as reasoning with certainty. This is where Automated Reasoning becomes increasingly important.

From LLMs and Generative AI to AI Agents and Neuro-Symbolic AI, the next generation of intelligent systems needs more than neural networks alone. It needs the ability to represent knowledge explicitly, search through possible solutions, enforce constraints, verify conclusions, prove correctness, and make decisions according to well-defined rules. Automated Reasoning provides many of these capabilities.

This course takes you on a journey from Logic, Knowledge Representation, Search, Constraints, and Planning to Theorem Proving, SAT/SMT Solving, Symbolic Execution, and Formal Verification — and finally connects these ideas to real-world systems such as CPU verification at Intel and spacecraft software verification at NASA.

You will not learn Automated Reasoning as a collection of complicated mathematical formulas or isolated algorithms. Instead, you will learn to understand why these technologies exist, what problems they were created to solve, how they fit together, and why they are becoming increasingly relevant to Generative AI, AI Agents, and the emerging Neuro-Symbolic AI paradigm.

Understand the Reasoning Behind AI — Not Just the Algorithms

Much of modern AI is built around neural networks. LLMs can learn patterns from enormous amounts of data and generate remarkably sophisticated responses. However, neural generation has an important limitation: an answer can be convincing without necessarily being guaranteed to be correct.

This creates a fundamental question: What happens when AI needs to do more than generate — when it needs to reason, plan, verify, and act reliably?

Automated Reasoning offers a different perspective. Instead of asking only “What answer is likely?”, we can ask:

What knowledge do we have?

What rules must hold?

What possibilities are allowed?

What constraints must be satisfied?

Can this conclusion be proven?

Can this program be guaranteed to behave correctly?

These ideas have existed for decades in logic, theorem proving, constraint solving, planning, and formal verification. What is changing today is their relationship with modern neural AI.

The emerging Neuro-Symbolic AI paradigm seeks to combine the strengths of both worlds:

• Neural AI provides perception, language understanding, learning, and flexible generation.

• Symbolic AI and Automated Reasoning provide explicit knowledge, structured reasoning, constraints, search, planning, and verification.

The result is not simply a larger language model. It is a different way of thinking about intelligent systems: neural models generate possibilities; symbolic systems can constrain, reason about, search through, and verify those possibilities.