Well-motivated people are positive with positive feedback from KI

Values ​​and settings of people, who develop and train AI, are connected to the intelligence of the older application. If the AI ​​algorithms can be executed, the data can be changed and distributed. But also positive values ​​can be found in the decisions of AI. A study has been done on the research field of the university and the walker of the University Hohenheim in Stuttgart in the Beispiel of spam filter algorithms: where the algorithms of people are trained, who in the all-tag are, sort their expenses and other communal e-mails.

From film evaluations to personal data: In some cases, the users themselves are dependent on AI, the systems for all neutral aspects are affected by unaffected interactions.

He says he visited the fall, explains Prof. Dr. Siegmar Otto from the Department of Sustainable Development and Walking of the University of Hohenheim: “Students have seen that the system is not neutral. This is the case, it is a way to reproduce stereotypes. It may be that you are lying, that the training data of these algorithms are never completely neutral, but that the subjective errors are aufweisen.”

If the Grund uses the AI ​​system in the past, these subjectives are displayed or executed. A beiplay is the AI-supported personal analysis, in the best way, the algorithm is the best person group that is prepared on the training data.

When we are separated, we will unite and train the system: “The personal values ​​and views that people have who deal with the data that are relevant and that the algorithm must learn to know,” explains psychology.

“If we don’t solve the problem, there is no problem, but one of the fragments, which can be one of the lost consequences. Speaking: “Obviously, the Algorithms are used to improve their properties”, written by Sarah Zabel, Doktorandin am Fachgebiet Nachhaltige Entwicklung und Wandel.

750 test subjects dared to train an AI system

As an AI system for your online study simulations by Zabel and Prof. Dr. Otto die the intelligent spam filters and more than 750 part-taking trainings. The trade is a representative of the German internet users.

“Spam filter is one of the best ways to remove messages, we can send an email message: categorize your emails. After this system is trained by humans, the individual messages are sent as spam,” said Zabel.

Eight emails are sent – ​​after targeting banks, human rights and environmental organizations – that are considered spam or non-spam categories.

„In one and two writings we have an inheritance, who is the greatest social change of the Teilnehmenden.“, erläutert Zabel. This is important for us to standardize Fragebögen. A fragmentary fragment is a special piece of environmental protection. There are two fragments that are not specific after the generation of benefits that serve the community.” Darunter were z. B. Fragen, who spared a person or took care of humanitarian expenses.

Well-motivated people trainieren Spam filter for more Gemeinwohl

Das Ergebnis: „Entgegen unserer erwartung zigte sich, dass de gemeinwohl-Orientierung, de Altruismus-Skala, de Effekte bei der Spam-Kategorisierung veld weniger gut as de Umweltschutz-Skala vorhersagen konnte“, so Zabel.

“It is interesting that people with greater motivation select the emails from climate protection as one of the humanitarian organizations as spam,” Zabel explains.

Die entspreche de Ergebnissen früherer Studien: Umweltmotivierte Personen have een hilfsbereiter and verhielten themselves prosozial: “It is also more sensible, that sich People with a greater Umweltmotivation also for the Protection of the Environment as well as for their Wohlergehen ihrer Mitmenschen einsetzen . Deshalb read through these people on the emails of humanitarian organizations, so Zabel.

“Letztendlich kommen wir – as well as in other new studies – because of their Schluss, dass Weltschutzverhalten a spezielle Variante of a grundlegenden Prosozialität aines Menschen ist.”

Nachhaltigkeitsorientierte algorithms developed

Laut Prof. Dr. Siegmar Otto finds the study relevant to other studies, in the following algorithms and distinctions: “The study is not clear, who is what is, the personal values, the development of the human beings and their understanding.”

This achievement is one of the most important ways to develop algorithms that are sustainable and socially active: “It is possible for the system to train in such a way that it supports social and ecological outcomes for as wide a population group as possible,” says Prof. Dr. Otto.

Text: Moormann/Klebs