The AIMA algorithms, ready to leave the textbook

What it does

aima-python is the companion code for Artificial Intelligence: A Modern Approach, the Russell and Norvig textbook. It implements the book’s pseudocode algorithms in Python 3.7+, covering search, games, constraint satisfaction, logic, and planning. Each topic gets a module with runnable implementations, explanatory Jupyter notebooks, and lightweight tests.

The interesting bit

The project updated for the book’s 4th edition by inverting the usual code-notebook relationship: the .py files are auto-generated from Jupyter notebooks, which serve as the primary documentation. It also modernized the codebase with type hints, dataclasses, and f-strings.

Key highlights

• Broad coverage of classical AI: A*, minimax, alpha-beta, CSP solvers, propositional and first-order logic, SATPlan, and more.
• Notebooks are the source of truth; readable .py files are generated from them.
• Uses modern Python 3.7 features like type hints, dataclasses, and f-strings.
• Includes a public algorithm index tracking implementation status, tests, and notebook coverage.
• Runs in a browser via Binder without local setup.

Caveats

• The algorithm index is still being updated for the 4th edition, and some entries—such as HybridWumpusAgent—lack any implementation, tests, or notebooks.
• Datasets live in a separate aima-data submodule that the README warns can take a while to fetch.
• Algorithms marked with an asterisk are noted as “not fully implemented.”

Verdict

Students and self-learners working through AIMA will find this indispensable; if you aren’t following the book or an AI course, it’s well-organized reference material you can largely find in dedicated libraries.