What Makes an AI Agent

IBUKI views an AI agent as having two parts: its boundry and its inside. The boundry connects an agent with its exterior and consists of its sensors and affectors some of which are priviledged in the sense that they are the primary means of communication. The inside has two parts: its cognative structure, which contains the parts of the agent that provide information and make behavioral decisions; and, its physics, which holds the agent together and constrains its physical configuration.

Building an AI Agent: The Boundry

One thing that makes an agent an individual is the ability (in principle) to determine what is 'in' it and what is 'outside'. That is, we can imagine it has a boundary. What makes it an autonomous individual is that its understanding of it external world can only be the result of information crossing its boundry via its sensors. In fact, one way to define a sensor is that it is any part of an individual that is used to understand its world. The ability to take sensor/perceptual data and turn it into cognative/semantic data is a talent our AI agent needs.

Where is the Boundry

The boundry of an individual is not a physical part of that individual but is part of a 'mental model' that 'contains' the individual that separates it from its exterior. The NEWFOL framework has a clean way of understanding thisy   find out more >>

Building an AI Agent: The Inside

Using th words of an old distinction, IBUKI divides the inside of an AI agent into its 'body' and 'mind'. mean those parts o    find out more >>

The Body

The body of an agent is on the inside and is like the skeleton or stomach of a biological agent. The body might consist of physical hardware like a computer or a camera but it may also be software. In all cases the body constrains the struture of the agent to conform to a pattern which NEWFOL refers as its 'physics'.    find out more >>

The Mind

By the 'mind' of an agent we mean both the parts that can think about the world and has goals and intensions (these correspond to what we usually think of as happening in the brain) and the parts that interpret the inputs from our sensors and decide what our affector are supposed to do. A thinking mind is modeled in NEWFOL as an IBML type called a [context]. [context]s are the most important data structure understood by the NEWFOL framework and, although can be used in more restricted ways, their use in building AI agents compares with neurons in biological agents.    find out more >>

Meet Fred - a friendly AI agent

Fred is an AI agent that demontrates the funtionality of an AI agent built using the NEWFOL framework. Fred is available by clicking here.