Examination of Artificial Intelligence

12. 10. 2020 Monday / By: Robert Denes / Generic / Exact time: BST / Print this page

Daniele Quercia and a research team used algorithms and the psychological concept of “integrated complexity” to find a possible solution.

Artificial intelligence (MI or AI - from the English Artificial Intelligence) is called intelligence manifested by a machine, program, or artificially created consciousness. The term is most often associated with computers. In common parlance, it is used in several separate meanings:

  • The artificially created object should be able to respond to environmental influences without constant human intervention (automation) - the simple software agent is such;

  • The artificially created object should be able to behave similarly to an organism with natural intelligence, even if there is a different mechanism behind the same behavior (TI simulation - in this sense we can talk about, for example, the “intelligence” of machine-controlled characters in computer games);

  • Finally, the artificially created object should be able to change its behavior in a practical and repeatable way (learning) - the latter meaning is what has come to the fore in modern MI research and is currently best identified by the concept of MI.

    Social media has been recently found to exacerbate political polarisation and violence, and current technological solutions have mostly focused on detecting and punishing those misbehaving (terminating the accounts of those engaging in bullying or hate speech) rather than celebrating those with healthy habits (the silent majority).

    To tackle that challenge, we have recently developed algorithms that automatically promote the healthy circulation of valuable information.

    These algorithms are based on the psychological concept of Integrated Complexity. This refers to a person’s ability to bridge opposing points of view. Recent work in Social Psychology has found Integrated Complexity to be related to the person’s use of language (for example, to the use of figurative expressions like “on the other hand”). Based on that literature, we trained machine learning algorithms that are able to spot markers of high Integrative Complexity in social media posts. For example, “Death penalty is definitely appropriate to punish all murderers. Society is better off without this type of criminals” is a sentence with low Integrative Complexity: it states one, non-negotiable point of view. Instead, the statement “Death penalty is an understandable attempt to right a wrong, but it does so with a similar wrong action. We should think about aggravating punishments for brutal crimes, but without inflicting physical harm to anyone” has higher Integrative Complexity because it recognizes the reasons of two opposing views and attempts to find a...read more

    Via Link




    Canning Town, Barking Road
    London E13 8EQ
    United Kingdom