The Delta Normal AGI Pieter ter Doest, MSc. PNF7, The Netherlands pieter.ter.doest@pnf7.nl www.pnf7.nl February 27, 2022 Abstract. In the quest for an Artificial General Intelligence (AGI) this paper presents a proposal for a symbol-based narrow AGI that uses a problem-driven mechanism within a certain domain. Using a small set of seeded ontology roots, simplified sentences can be constructed with surprising characteristics. Problem solving graphs with a limited depth are combined to form larger graphs. Keywords: AGI · COFO · Problem Solving · Combinatorial processes 1 Introduction Problems are universal. Each moment of the day we encounter problems, al- though we are not aware of this all the time. Washing the car can be seen as a problem. The same goes for driving to work, or picking up a fork. We can look at all these activities as presenting problems. But also a dialogue in fact is a combination of mini problem-solving dialogues in order to solve a particular problem. A central point in this paper is the thought that ’problem solving dia- logues’ (PSD) are central in conversations, navigation, problem solving or in any knowledge based interaction. Even a unit of information itself can be seen as a PSD. When a problem is getting more complex, then more PSD’s are needed. 2 Definitions First we need to agree on some definitions: 2.1 The Topic-region A topic-region is a small part of a larger graph. We can apply a modified version of sampling theory: From a given point there are enough topic regions so that a Turing Test is satisfied. Several topic-regions together form a topic. An example of a topic = ’talking about lunch’. Example of a topic-region = ’the jam on my toast’. A topic-region should be small enough so that the AGI will have a limited set of solutions. We want to turn complexity into quantity: a complex convoluted graph can be