Classic Cognition aims to track the progress of commoditized (and affordable) deep learning AI on classic problems. By analogizing your business challenges to these 'games' or cognitive tasks, you can determine the appropriate level of AI or automation for your needs. This comprehensive resource provides insights into various domains where deep learning is applied, categorizing tasks to help you navigate the current state of AI technology.
| Classic Problem/Game | Category | Deep Learning Applicability | Current State of Commoditized AI | Sample LLM Prompt | Details |
|---|---|---|---|---|---|
| Simon | Memory and Pattern Recognition | High | AI can predict and replicate patterns. | N/A | View Details |
| Concentration (Memory card game) | Memory and Pattern Recognition | Moderate | AI can memorize positions but lacks human-like strategies. | N/A | View Details |
| Tetris | Spatial Reasoning and Navigation | High | AI can play at superhuman levels. | N/A | View Details |
| Portal | Spatial Reasoning and Navigation | Low to Moderate | AI struggles with complex 3D navigation. | N/A | View Details |
| Chess | Strategy and Planning | Very High | AI surpasses human experts. | N/A | View Details |
| Civilization | Strategy and Planning | Moderate | AI manages resources but lacks depth. | N/A | View Details |
| Sudoku | Problem-Solving and Logic | High | AI solves puzzles efficiently. | "Solve this Sudoku puzzle." | View Details |
| Escape Rooms | Problem-Solving and Logic | Low | AI struggles with open-ended puzzles. | N/A | View Details |
| Where's Waldo? | Attention and Reflexes | High | AI can detect objects in images. | "Find Waldo in this image." | View Details |
| First-Person Shooter Games | Attention and Reflexes | Moderate | AI can aim but lacks strategic depth. | N/A | View Details |
| Scrabble | Language and Verbal Skills | High | AI can generate high-scoring words. | "What is the best word with these letters?" | View Details |
| 20 Questions | Language and Verbal Skills | Moderate | AI struggles with contextual understanding. | "Guess the object I'm thinking of." | View Details |
| Blackjack | Mathematical and Numerical Skills | High | AI plays optimally using probability. | "Should I hit or stand with this hand?" | View Details |
| Set | Mathematical and Numerical Skills | High | AI can identify sets quickly. | "Find all sets in this group of cards." | View Details |
| Poker | Decision Making and Risk Assessment | Very High | AI outperforms humans in strategy. | "What is the optimal move given this hand?" | View Details |
| Stock Market Simulations | Decision Making and Risk Assessment | Moderate | AI predicts trends but with limitations. | "Should I buy or sell this stock?" | View Details |
| Pictionary | Creativity and Imagination | Moderate | AI can recognize drawings. | "What does this drawing represent?" | View Details |
| Minecraft | Creativity and Imagination | Low to Moderate | AI can perform tasks but lacks creativity. | N/A | View Details |
| Dance Dance Revolution | Physical-Cognitive Coordination | Low | AI lacks physical embodiment. | N/A | View Details |
| Operation (board game) | Physical-Cognitive Coordination | Low | AI can't perform fine motor tasks. | N/A | View Details |
| The Sims | Multitasking and Resource Management | Moderate | AI manages tasks but lacks spontaneity. | N/A | View Details |
| Overcooked | Multitasking and Resource Management | Moderate | AI can coordinate but struggles under pressure. | N/A | View Details |
| Minesweeper | Pattern Completion and Prediction | High | AI solves puzzles efficiently. | "What is the next safe move?" | View Details |
| Wordle | Pattern Completion and Prediction | High | AI suggests optimal guesses. | "What should my next guess be?" | View Details |
| Among Us | Social Cognition and Deduction | Low | AI struggles with deception and social cues. | N/A | View Details |
| Charades | Social Cognition and Deduction | Low | AI lacks physical interaction. | N/A | View Details |
| Typing Games | Speed and Efficiency | Low | AI types faster than humans. | N/A | View Details |
| Speed Chess | Speed and Efficiency | Very High | AI performs exceptionally under time constraints. | N/A | View Details |
| Super Mario Bros. | Platformers and Motor Skills | High | AI can complete levels efficiently. | N/A | View Details |
| Mirror's Edge | Platformers and Motor Skills | Low to Moderate | AI struggles with complex movements. | N/A | View Details |
| Gran Turismo | Driving and Racing Games | High | AI drives efficiently on known tracks. | N/A | View Details |
| Mario Kart | Driving and Racing Games | Moderate | AI handles basic driving but struggles with items. | N/A | View Details |
| The Legend of Zelda | Puzzle and Adventure Games | Low to Moderate | AI struggles with open-world exploration. | N/A | View Details |
| Professor Layton Series | Puzzle and Adventure Games | Moderate | AI can solve individual puzzles. | "Solve this logic puzzle." | View Details |
| Boggle | Word and Language Games | High | AI finds all possible words quickly. | "List all words from this grid." | View Details |
| Crossword Puzzles | Word and Language Games | Moderate | AI solves clues with varying success. | "What is a 5-letter word for 'ocean mammal'?" | View Details |
| Go | Classic Board Games | Very High | AI exceeds human expertise. | N/A | View Details |
| Monopoly | Classic Board Games | Moderate | AI can make optimal financial decisions. | N/A | View Details |
| StarCraft | Real-Time Strategy Games | Very High | AI competes at professional levels. | N/A | View Details |
| Age of Empires | Real-Time Strategy Games | High | AI manages resources effectively. | N/A | View Details |
| Meal Planning and Cooking | Common Cognitive Tasks | Moderate | AI can generate recipes and schedules. | "Plan a week's meals under $50." | View Details |
| Navigation | Common Cognitive Tasks | High | AI provides optimal routes using GPS data. | "Find the fastest route to this address." | View Details |
The game Simon tests memory through repeating increasingly complex color and sound sequences. Deep learning models can predict and replicate these patterns with high accuracy.
Relevant Article: "Pattern Recognition Using Deep Learning"
Codebase/Project: Pattern Recognition GitHub Repository
In Concentration, players match pairs of face-down cards. AI can memorize card positions but lacks the adaptive memory strategies humans use.
Relevant Article: "Memory-Augmented Neural Networks"
Codebase/Project: Memory Networks by Facebook
Tetris involves rotating and placing falling blocks to complete lines. AI using deep reinforcement learning can achieve superhuman performance.
Relevant Article: "Deep Learning for Tetris"
Codebase/Project: Tetris AI GitHub Repository
Portal requires solving physics-based puzzles in a 3D environment. AI struggles with the complex spatial reasoning and physics involved.
Relevant Article: "Learning to Navigate in Complex Environments"
Codebase/Project: AirSim by Microsoft
Deep learning has revolutionized chess AI, with models like AlphaZero learning to play at superhuman levels through self-play.
Relevant Article: "Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm"
Codebase/Project: Leela Chess Zero
Civilization requires managing and expanding an empire over time. AI can handle resource management but often lacks the nuanced long-term planning of human players.
Relevant Article: "Strategic Planning in Strategy Games"
Codebase/Project: Civilization V AI Modding Community
Sudoku involves filling a grid with numbers following specific rules. AI can solve Sudoku puzzles efficiently using constraint satisfaction and deep learning techniques.
Relevant Article: "A Deep Learning Approach to Sudoku Solving"
Codebase/Project: Sudoku Solver with Deep Learning
Escape Rooms require solving a series of puzzles to "escape" a themed room. AI struggles with the open-ended and physical nature of these tasks.
Relevant Article: "Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning"
Codebase/Project: TextWorld by Microsoft
"Where's Waldo?" challenges players to find a character in a complex scene. AI can use object detection and image recognition to locate Waldo efficiently.
Relevant Article: "Building a Waldo-Finding AI"
Codebase/Project: Where's Waldo AI Project
In FPS games, players need quick reflexes and aiming skills. AI can aim and shoot with precision but often lacks tactical and strategic understanding.
Relevant Article: "Reinforcement Learning in First-Person Shooter Games"
Codebase/Project: ViZDoom Reinforcement Learning Platform
Scrabble involves forming words on a board for points. AI can generate high-scoring words using extensive vocabulary databases and optimal placement algorithms.
Relevant Article: "Artificial Intelligence for Scrabble"
Codebase/Project: Scrabble Bot Project
The game involves guessing an object through yes/no questions. AI struggles with inference and contextual understanding required for this task.
Relevant Article: "Neural Question Generation for 20 Questions Game"
Codebase/Project: Hedwig: Deep Learning Models for Question Answering
Blackjack requires quick addition and probability assessment. AI can play optimally using card counting and reinforcement learning strategies.
Relevant Article: "Deep Reinforcement Learning for Blackjack"
Codebase/Project: Blackjack AI with Reinforcement Learning
Set is a card game where players find patterns among cards with various attributes. AI can identify sets quickly using image recognition and pattern matching.
Relevant Article: "Neural Networks for Card Game Pattern Recognition"
Codebase/Project: Set Game Solver
Poker involves probability, bluffing, and risk management. AI like Libratus and Pluribus have outperformed professional players using deep reinforcement learning.
Relevant Article: "Superhuman AI for Multiplayer Poker"
Codebase/Project: Libratus (Refer to publications)
Stock market simulations require making investment decisions based on market analysis. AI predicts trends using historical data but is limited by market unpredictability.
Relevant Article: "Deep Reinforcement Learning for Stock Trading"
Codebase/Project: Reinforcement Learning for Stock Prediction
Pictionary involves conveying words through drawings. AI can recognize simple drawings using convolutional neural networks.
Relevant Article: "Sketch-RNN: A Generative Model for Drawing"
Codebase/Project: Quick, Draw! Dataset
Minecraft allows building and crafting in an open world. AI can perform tasks but lacks the creativity and planning skills of human players.
Relevant Article: "Towards AI in Minecraft: State of the Art"
Codebase/Project: Project Malmo by Microsoft
Dance Dance Revolution requires stepping on arrows in time with music. AI lacks physical embodiment to perform this task.
Relevant Article: N/A
Codebase/Project: N/A
Operation tests fine motor skills by removing pieces without touching the edges. AI cannot replicate the physical dexterity required.
Relevant Article: N/A
Codebase/Project: N/A
The Sims involves managing virtual characters' lives and environments. AI can handle routine tasks but lacks spontaneous behavior and emotional understanding.
Relevant Article: "AI for Simulated Characters"
Codebase/Project: N/A
Overcooked is a cooperative cooking game requiring coordination and time management. AI can perform individual tasks but struggles with team coordination under pressure.
Relevant Article: "Learning to Collaborate in Multi-Agent Systems"
Codebase/Project: Overcooked-AI
Minesweeper involves deducing mine locations from numerical clues. AI can solve puzzles using probabilistic models and constraint satisfaction algorithms.
Relevant Article: "Reinforcement Learning for Minesweeper"
Codebase/Project: DeepMines Project
Wordle challenges players to guess a five-letter word with limited attempts. AI can suggest optimal guesses using frequency analysis and pattern matching.
Relevant Article: N/A (Due to the game's recent popularity)
Codebase/Project: Wordle Solver
Among Us requires identifying impostors through social deduction. AI struggles with interpreting social cues and deception.
Relevant Article: "Social Deduction with AI Agents"
Codebase/Project: N/A
Charades involves guessing words or phrases from actions without speaking. AI lacks the ability to perform or interpret physical gestures effectively.
Relevant Article: "Deep Learning for Gesture Recognition"
Codebase/Project: Hand Gesture Recognition
Typing games aim to improve typing speed and accuracy. AI can type faster than humans but lacks practical applications in this context.
Relevant Article: N/A
Codebase/Project: N/A
Speed Chess requires quick decision-making under time constraints. AI excels in this domain due to rapid computation capabilities.
Relevant Article: "Real-Time Decision Making in Games"
Codebase/Project: Stockfish Chess Engine
Super Mario Bros. involves navigating obstacles and timing jumps. AI can complete levels using reinforcement learning and neural networks.
Relevant Article: "Reinforcement Learning for Mario AI"
Codebase/Project: Gym Super Mario Bros
Mirror's Edge is a first-person parkour game requiring spatial awareness. AI struggles with the complex movements and environmental interactions.
Relevant Article: N/A
Codebase/Project: N/A
Gran Turismo is a realistic driving simulator. AI can drive efficiently on known tracks but may struggle with unexpected conditions.
Relevant Article: "Deep Reinforcement Learning in Racing Games"
Codebase/Project: Udacity Self-Driving Car Simulator
Mario Kart introduces dynamic elements like power-ups and obstacles. AI handles basic driving but struggles with strategic use of items.
Relevant Article: "Reinforcement Learning in Mario Kart"
Codebase/Project: Learning to Drive in 5 Minutes
The Legend of Zelda combines puzzle-solving, exploration, and combat. AI struggles with open-world exploration and dynamic puzzle-solving.
Relevant Article: "Generating Complex Levels in The Legend of Zelda"
Codebase/Project: Zelda AI Project
Professor Layton games involve solving various logic puzzles within a narrative. AI can solve individual puzzles but lacks narrative understanding.
Relevant Article: N/A
Codebase/Project: N/A
Boggle requires finding words in a grid of letters under time pressure. AI can find all possible words using dictionary lookups and grid traversal algorithms.
Relevant Article: N/A
Codebase/Project: Boggle Solver
Crossword puzzles involve filling in words based on clues. AI can solve clues using natural language processing but may struggle with nuanced hints.
Relevant Article: "Solving Crossword Puzzles via ProBERT"
Codebase/Project: ProBERT Project
Go is an ancient board game of territory control and long-term strategy. AI like AlphaGo has exceeded human expertise using deep reinforcement learning.
Relevant Article: "Mastering the Game of Go with Deep Neural Networks and Tree Search"
Codebase/Project: Leela Zero
Monopoly involves property trading and financial decision-making. AI can make optimal decisions but may not account for human negotiation tactics.
Relevant Article: N/A
Codebase/Project: Monopoly AI Project
StarCraft requires managing resources, building armies, and tactical combat. AI like AlphaStar competes at professional levels using deep learning.
Relevant Article: "Grandmaster Level in StarCraft II Using Multi-Agent Reinforcement Learning"
Codebase/Project: PySC2 by DeepMind
Age of Empires is a historical strategy game with resource management and warfare. AI can manage resources but may lack complex strategic planning.
Relevant Article: N/A
Codebase/Project: OpenAge Project
Meal planning involves recipe following, timing, and multitasking. AI can generate recipes and schedules but may not account for personal preferences.
Relevant Article: "DeepChef: AI for Recipe Generation"
Codebase/Project: Microsoft Nutrition API
Navigation requires using maps or GPS to find routes in unfamiliar areas. AI provides optimal routes using real-time data and deep learning algorithms.
Relevant Article: "Deep Learning for Shortest Path Routing"
Codebase/Project: Google Research GPS
Classic Cognition is an ongoing project. We encourage the community to contribute by updating information, adding new tasks, or correcting inaccuracies. If you'd like to contribute, please submit a Pull Request (PR) on our Codeberg git repository.
Main Project Repository: ClassicCognition.com git repo
Stay informed about the state of AI and how it can be applied to solve classic and modern problems. By understanding the capabilities and limitations of deep learning, you can make informed decisions about integrating AI into your business or projects.