8.2 Artifical Intelligence

• Term refers to the cognitive abilities and knowledge stocks of a person at any one time

• Science that develops new solutions to technical problems based on an understanding of nature

e.g.

• swarm intelligence: swarm is created when individuals interact with one another directly without central control -> overall increase in efficiency

• Chemical messengers from individuals that transmit information

• type of swarm intelligence (bionics)

• used in routing communication networks

• principle taken from nature: ants lay a trail of pheromones -> fastest way back to nest = ant that took the shortest way -> marked the route with pheromones on both ways (double pheromones) -> other ants learn and take the shortest route in future

(Windt, 2006)

• type of swarm intelligence

• outsourcing of task usually done within organisation

• outsourcing to the crowd -> group of volunteers on the Web; e.g. Wikipedia

  • advantages of crowdsourcing: use of creativity of many people -> individuals are motivated b/c ideas are heard

  • incentives through prizes

• crowdsourcing plattform -> managed by artifical intelligence

(Kiel, 2015)

• Computer systems that imitate human intelligence

• Applies when the following is fullfilled:

  • 1) ability to learn;

  • 2) ability to deal with uncertainty and probabilities

  • 3) ability to solve problems in an abstract way

(Specht, 2018)

• comparable performance to human brain

• Consciousness, environmental sensations, feelings

• today’s systems superior to humans: Alpha Go (chess)

• AI learns from innumerable people; however: weak in dealing with complexity, ambiguity and contradiction

• Baldwin (2019): AI is not a baby farmer that will fully grow; it is like tractor, a tool

• used to determine if a computer can think

• test based on two interlocturs communication via a screen -> one is a chatbot (text-based dialogue system)

• if test person cannot determine which of the two interlocutors was the computer, then test passed = strong AI

(Eberl, 2016)

• Englishman Alan Mathison Turing

• Deciphered Enigma machine used to encrypt German messages during WWII

• basis for AI

• systems fed with examples from which they recognis patterns and regularities -> then make generalisations

• Machine learning = deep learning

• needs large amounts of training data and powerful computers

• type of machine learning

• software programmes that mimic brain functions

• virtual interconnected neurons are arranged in several layers -> signals passed on

• computer supplied with large amount of training data -> i.e. photos labelled “dog” and “no dog” -> structures and patterns analysed layer by layer with increasing complexity

  • first layer: pixels

  • second layer: connection of pixels to lines

  • third layer: horizontal and vertical lines usw.

  • last layer: result “dog”

Computer vision: extensive Image recognition -> used in autonomous vehicles to evaluate camera images

Biometrics: recognition of person based on individual biological characteristics (i.e. passport)

Speech recognition: ability to record texts on PC (Siri)

Natural language processing: text and language interpreted -> translation programmes improved (i.e. Deepl)

Sentiment detection: feelings recognised in text and speech -> use in customer service

robots: artifical humans

(Specht 2018)

• popular through Russian-American biochemist Isaac Asimov in his sci-fi novels

• Three principles of robotics:

  • A robot may not injure a human being or cause harm through inactivity.

  • A robot must obey the orders of a human, unless such orders are contrary to the

    first law.

  • A robot must protect its own existence as long as this protection does not conflict

    with the first or second law.

• robots that move in the manner of a human

• e.g. catch ball -> real-time prediction how object will fly; how arm and hand are in correct place etc.

• challenge to map the human body with muscles, ligaments and tendons -> large number of sensors and immense amount of data

Industrial robots =

“universally applicable automatic machines with several axes, whose movements are freely programmable (i.e., without mechanical or human intervention) with regard to the sequence of movements and paths or angles, and are sensor-guided if necessary.

They can be equipped with grippers, tools, or other manufacturing equipment and can perform handling and/or manufacturing tasks” (Länger, 2020).”