How does artificial intelligence work?

[tongfkymm44]

How can we describe the functioning of artificial intelligence? An AI is only as good as its technical representation of knowledge . In this sense, there are two basic methodological approaches: the symbol processing approach and the neural approach.

In the symbol processing approach to AI, knowledge is represented through symbols and realized through the manipulation of symbols. Symbolic AI focuses on processing information in a “top-down” manner and operates using symbols, abstract connections, and logical conclusions.
In the neural approach to AI, knowledge is represented by artificial neurons and their interconnections. Neural AI focuses on bottom-up information processing and simulates the functions of individual artificial neurons that are grouped into larger neural networks. These artificial neural networks work together to perform specific tasks.
IONOS AI Model Hub
Your gateway to a secure multimodal AI platform
A platform for the most powerful AI models
Fair and transparent token-based pricing
No vendor lock-in thanks to open source
Symbolic AI
Symbolic AI is considered the classical approach to artificial intelligence . It is based veterinary email database on the premise that human thought can be reconstructed from a higher logical-conceptual level , independently of specific empirical details (top-down approach). In this approach, knowledge is represented by abstract symbols, including written and spoken language. Machines learn to recognize, understand, and use these symbols through algorithms. The intelligent system acquires its information from expert systems , which are systems designed to store and apply specialized knowledge in a specific domain.

Classic applications of symbolic AI focus on word processing , speech recognition , and other logical disciplines such as chess . Symbolic AI is based on established rules and can solve increasingly complex problems as computers' computing power increases. For example, IBM's Deep Blue beat the then world chess champion Garry Kasparov in 1996 with the help of symbolic AI.

Neural AI
In 1986, Geoffrey Hinton and two of his colleagues revived neural AI research and with it the field of artificial intelligence research. By perfecting the backpropagation algorithm , they created the basis for deep learning , which almost all AI works with today. Thanks to this learning algorithm, deep neural networks can constantly learn and grow independently, thus overcoming challenges that symbolic AI fails to tackle.

Like the human brain, neural AI divides knowledge into smaller functional units, known as artificial neurons, which are connected in a network forming increasingly larger groups (bottom-up approach). The result is an artificial neural network with multiple branches . Unlike symbolic AI, the neural network is “trained” using sensorimotor data, as happens in robotics. Through these experiences, AI generates constantly growing knowledge. Herein lies the great innovation: although training can take time, the system is ultimately capable of learning autonomously.