Putting emotion into a machine may seem elusive, unimaginable and maybe even impossible (but it isn't, as we've done it). Trying to do that feels close to putting a 'Ghost in the Machine'. But that is precisely what we went for because from our perspective you need a 'Ghost' to have understanding, insight, moral values even. Emotion becomes the driver for those things, as they are in humans.

Our moral and ethical rules are all fundamentally based on emotions. If you want a machine to 'understand' those rules, to act by them, it needs the emotional values that make those rules perceptible.

Machine Consciousness is a special area in the field of Artificial Intelligence research. Not much progress has been made so far (outside of MIND|CONSTRUCT), although there are many books and papers written on the subject (including the paper by Hans Peter, linked below). There appears to be some consensus in the field, that Machine Consciousness might be the required property for both General Intelligence and safe intelligent machines.

Having technology like the ASTRID system brings great insights into these questions. We can observe how emotional values spread through the ASTRID brain, similar to 'spreading activation' in a human brain. Machine Consciousness might be just around the corner.

As far as we know, nobody in the Artificial Intelligence community has deemed Emotion fundamental for reasoning. There are projects that try to build 'affect' into robotics, trying to make robots appear (not actually be) friendly and understanding. Of course, this has nothing to do with actual 'understanding'.

However, in the field of Neural Science, research by several well-known scientists (like e.g., Antonio Damasio) has shown that emotion might be instrumental in cognition, and even crucial for the ability to interact with the world. From this, it appears that research into the brain's mechanisms for emotion should be high on the agenda of every AI researcher and developer.

Nothing seems further from reality, though.

When Hans Peter started the project, over a decade ago, he was convinced that emotion played a major role in cognition. When he modelled the ASTRID system, it was clear from the start; emotion was a fundamental property in the model.

Over the years of development, this has shown to be an incredible important decision, as various arising questions about the model showed to be no issue at all because of the functional presence of emotion in the model.

In the ASTRID-system, emotion is implemented through the Emotional Bias Response Model (EBRM), which basically validates any resulting inference against a dynamic emotional framework.