This enables developments for efficient, sustainable applications. While the industrial revolution replaced muscle power, the "digital revolution" with AI is characterized by the transfer of cognitive services to a "machine" that becomes smarter from experience. In addition to the risk to cyber security, the downside is the disappearance of human labor, especially simple labor, and falling tax revenues, as "intelligent machines" are tax-exempt. On the other hand, new fields of activity are emerging. International studies support this optimism.

What is AI
AI uses a combination of planning, searching, optimizing, logical processing and recognition via approximation methods to investigate and solve intelligent automation with machine learning by using fast computer networks with high computing power. Human action structures are simulated. Such fail-safe, error-avoiding computers and networks can independently process problems and suggest better decisions. They are not intelligent in the sense of a mixture of logical thinking, empathy, tact and ethics.

AI is often defined as an implementation that simulates "intelligent behavior" using special algorithmic programs (e.g. data analysis programs, computer games). Such "weak" AI is contrasted with "strong" AI. It provides support for difficult tasks that are characterized by uncertainties and probabilities. Its way of working often surpasses human behavior. "Strong" AI solutions require the cooperation of a wide variety of skills, programs and systems based on large amounts of data in order to create optimal solutions flexibly and reliably. The beginnings of AI date back to the 1940s.

Realization
An AI system consists of the functional groups shown in the picture (top right), whose components are arranged in a decentralized manner with varying degrees of power depending on the task. All of them are reliably networked at the highest data transmission speed. A high-performance computer acts as the control center, whose operational speed is realized alongside supercomputers using quantum-class devices and artificial, neural structures KNN with the ability of machine learning ML (machine or deep learning).

Quantum computers work with molecules, atoms and subatomic particles. Phenomena such as superposition and quantum entanglement become usable. The quantum bit qubit replaces the classic bit and can form zero, one or both at the same time. Extremely fast processing in minimal space becomes usable, the computing power of the qubits increases exponentially with their quantity (20 qubits can already occupy millions of bit states). The latest developments use the spin of an electron as a qubit. The associated solid consists of two semiconductor layers (such as silicon or germanium close to absolute zero). The free electron is directed by electric fields. Other approaches use the direction of current in a superconductor ring or the energy level in atoms or molecules. The qubit numbers achieved to date are 50-70.

KNN is modeled on the information architecture of the human brain. The corresponding computer system can recognize hidden correlations better than a human being. Using mathematical and stochastic methods, it transforms disorganized raw data (such as measurement results and photos) into structured information in order to derive correlations and trends. On the one hand, CNNs are simulated using software. On the other hand, there are neuromorphic systems consisting of electronic, network-compatible resistors with memory - the memristors (artificial word from "memory" and "resistor").

The design and location of the two other, functionally known shells in the diagram above depend on the task and spatial extent as well as the safety requirements.