The intelligence of computers or software, as opposed to the intellect of people or other creatures, is known as artificial intelligence (AI). It is a branch of computer science that focuses on creating and researching intelligent machines.
Artificial Intelligence is widely applied in government, industry, and academia. Advanced online search engines (like Google Search), recommendation engines (like YouTube, Amazon, and Netflix), speech recognition (like Google Assistant, Siri, and Alexa), self-driving cars (like Waymo), generative and artistic tools (like ChatGPT and AI art), and superhuman play and analysis in strategy games (like chess and Go) are a few high-profile applications.[1]
The first significant researcher in the topic he named "machine intelligence" was Alan Turing.[2] Machine learning To achieve the aforementioned objectives, AI research employs a wide range of methodologies.[b]
Artificial intelligence is capable of finding intelligent solutions to a wide range of challenges.[67] AI uses two very different types of search: local search and state space search.
State-space exploration State space search looks for a target state by sorting through a tree of potential states.[68] For instance, means-ends analysis is the process by which planning algorithms look through trees of objectives and subgoals in an effort to locate a route to a target goal.[69]
For most real-world issues, simple exhaustive searches [70] are rarely sufficient because the search space—that is, the number of places to explore—grows exponentially fast. This leads to an excessively slow or nonexistent search.Mathematical optimization is used in local search to solve problems. It starts with a hunch and gradually gets more accurate.[73]
Gradient descent is a sort of local search technique that minimizes a loss function by gradually optimizing a collection of numerical parameters. evolutionary computation is a different kind of local search that selects just the most suitable solutions to survive each generation in an effort to iteratively enhance a set of candidate solutions through "recombining" and "mutating" them.[75]
Swarm intelligence methods can be used to coordinate distributed search operations. Particle swarm optimization, which draws inspiration from bird flocking, and ant colony optimization, which draws inspiration from ant trails, are two well-known swarm methods used in search.[76]
An agent must make decisions based on partial or ambiguous information in order to solve numerous AI issues, including those involving robotics, learning, planning, reasoning, and vision. AI researchers have used techniques from probability theory and economics to create a variety of tools to address these issues.[85]
A very versatile tool, Bayesian networks[86] can be applied to a wide range of tasks, such as reasoning (by means of the Bayesian inference process).[88] learning (using the algorithm of expectation-maximization).
In order to help perception systems comprehend processes that happen over time, probabilistic algorithms can also be used to filter, predict, smooth, and provide explanations for streams of data (e.g., hidden Markov models or Kalman filters).[92].Classifiers (such as "if shiny then diamond") and controllers (such as "if diamond then pick up") are the two categories into which the most basic AI applications can be separated.
You must be logged in to post a comment.