AI projects can be categorized into seven different patterns. Each pattern follows its own objectives, development iterations, considerations, risks, and complexities.
1. Conversational and Human Interaction
- The Conversational and Human Interaction pattern focuses on machines and humans interacting with each other using conversational forms of interaction across communication platforms such as voice, text, and images.
- Examples of this pattern include Chatbots, voice assistants, content generation, sentiment, mood and intent analysis, and machine translation.
- The objective of this pattern is for machines to interact with humans the way that humans interact with each other using human language instead of machine language or software.
2. The Recognition Pattern
- The Recognition pattern enables machines to identify and understand real-world and unstructured data such as images, sound, video, handwriting, faces, and gestures.
- Examples of this pattern include identification of faces, features, sounds, items, objects, handwriting, and text.
3. Predictive Analytics and Decision Support Pattern
- Predictive Analytics and Decision Support is an AI pattern that uses ML and other cognitive approaches to understand how to use past or existing behavior, interaction, or data to predict future outcomes.
- The pattern lets you predict various possibilities when making a certain decision or changing a certain parameter. Knowing how your current decisions could impact future outcomes gives you clarity and allows you to make informed choices.
- This pattern can be used for dynamic or predictive pricing; predicting potential failure of equipment or situation or if a bridge would collapse if certain building materials were replaced.
- The pattern can also be used to improve situational awareness by analyzing data and predicting future risks. It also has great applications in cybersecurity and fraud detection.
4. Goal-Driven Systems Pattern
- This pattern uses reinforcement-learning for real-world games. The objective of this pattern is to find the optimal solution to a problem through trial and error.
- Machine learning has proven to be remarkably good at discovering “hidden rules” of a game and beating even the best human players at their own games, such as chess and Go, which are considered incredibly complex.’
- Example use cases of this pattern include Creating simulations using scenarios, playing games, optimizing resources such as money, equipment, time; iterative problem-solving, Robo-advising, Bidding and real-time auctions.
5. Hyper-Personalization Pattern
- This AI pattern uses machine learning to develop a unique profile of an individual. Profile changes and evolves over time based on the real-time information of the individual. This pattern helps in creating customer profiles to target custom products and services.
- Healthcare: Create a patient profile to recommend treatments for a patients’ specific ailments, record the patient’s responses to treatments, and adjust treatments based on the patient’s response to draft a personalized plan.
- Finance: Create personalized finance plans and different investment options for an individual based on their income, spending habits, and attitudes to risks.
- Education and Training: Create personalized training plans based on learner needs, prior knowledge, and learning goals.
- Law Enforcement and Security: Create behavior profiles that can be used for law enforcement and security.
6. Autonomous Systems Pattern
- Autonomous systems are physical and virtual systems that are able to accomplish a task, achieve a goal, interact with surroundings, and perform their objective with minimal or human involvement.
- The objective of autonomous systems pattern is to minimize the need for human labor or cognitive process. Some of the applications of autonomous systems are:
- Autonomous vehicles: These refer to any transportation systems that are able to travel without a human driver.
- Autonomous robots: They are robots that can operate autonomously. Potential applications are in the warehouses, manufacturing units, hospitals, and retail spaces.
- Autonomous software systems: These systems can run and operate processes and tasks without human intervention, such as running business processes and software systems, and autonomously connecting between different software systems.
- Difference between Autonomous and Automation: Automation helps minimize human labor, but it only handles repetitive, routine tasks. Autonomous tasks are intelligent, giving them the ability to handle dynamic and complex situations in environments that cannot be predicted.
7. Patterns and Anomalies Pattern
- Patterns are events or items that can be grouped together or follow a common logic. Anomalies are events or items that don’t fit into those specified patterns or common logic.
- This pattern has applications in Fraud detection analysis, Content summarization, Cybersecurity, and Predictive analysis involving time series data.