What Is An Intelligent System?

What is Intelligence

First, some agreement on what constitutes intelligence is necessary. Depending on the discipline different attributes are valued and therefore constitute intelligence.

  • Effective speakers select words considering the sounds, grammar and meaning
  • Musicians create and communicate based on sound affect, pitch and rhythm.
  • Mathematicians utilize relationships without actions and abstract complex ideas.
  • Physicists can visualize/change/recreate images independent of referencing physical objects
  • Athletes/dancers control part of or the entire body using motor skills to manipulate objects

A person’s intelligence within a discipline would be recognized by their mastery of the discipline’s attributes. An argument can be made that a machine can be intelligent by mastering one set or multiple sets of attributes.

Elements of Intelligence

When evaluating something as intangible as intelligence, what elements should be considered? It starts with reasoning processes that support extending observable facts/events to general beliefs and/or take general beliefs consider alternatives in coming to specific conclusions. For a baseball player it is baseball IQ; the musician it is a sense of timing or beat; physicist the thought experiment in visualizing the abstract. It includes learning by training, observing and experience which expands the understanding of the discipline. 

Depending on the discipline, listening, recalling experiences, duplicating physical activities and drawing conclusions by relating environmental cues are all techniques for expanding knowledge. These techniques make a decision or choose a process which resolves an issue. They must also consider their surroundings, sensing valid inputs and providing structure. Generating value and knowledge, machines must also have communications skills to understand inputs and document results.

A picture containing text

Description automatically generated

Machine Intelligence

Humans rely on recognizing patterns in applying their intelligence. Each pattern, learned or experienced, has its information and heuristics for a situation. Familiar or stressful situations are facilitated by quick thinking in applying the most familiar pattern. New or complex situations require more thoughtful thinking in perceiving differences, altering existing or developing new patterns. 

Machine intelligence is programmed with attributes focused on a specific discipline. This intelligence relies on taking general beliefs considering alternatives arriving at conclusions. This can include understanding that a mistake has been made and avoid it in the future. 

Intelligent machines evaluate inputs, data, by applying sets of rules. While humans search for patterns, machines learning relies on rules/algorithms. Two different techniques are used to develop learning in machines. Supervised learning uses sample data with matched outputs (labelled data) to teach machines to approximate outcomes for new inputs. Unsupervised learning does not rely on labeled outputs, the learning occurs by inferring the natural structure of a set of data points. A differentiator between machine intelligence vs. human intelligence is in situations with incomplete or distorted inputs humans can determine outcomes where intelligent machines can have inconsistent results.