Reading List

The Selfish Gene
The Psychopath Test: A Journey Through the Madness Industry
Bad Science
The Feynman Lectures on Physics
The Theory of Everything: The Origin and Fate of the Universe


ifknot's favorite books »

Thursday 3 September 2015

The Most Intelligent Robot in the World: A Theory of Robotic Motivation.



It was my first time at the Grimsby TechSpace Meetup tonight and what a lovely warm welcome from a very eclectic mix of folk. In amongst the hybrid vigour of ideas that ensued consequently I was witness to a small epiphany on the subject of artificial intelligence...





It happened somewhere in the middle of the debate about whether or not AI would come about as an intended consequence of deliberate, directed construction or would arise as an emergent phenomenon as an unintended consequence from a fertile vat of complexity? The following then, perhaps, isn't even a new idea but it was new to me and I liked it... a lot!

So.

Amidst the conversation about various justified beliefs about "artificial intelligence" arose the quietly expressed proposition that anything truly intelligent, artefact, or otherwise is entirely self interested and, therefore, can not be utilitarian - a true AI is nobody's tool.

For me this idea chimed with Maslow's hierarchy of needs and I engaged my iconoclast in his impassioned views about the fundamental drivers of human behaviour (food and sex) from where we moved onto the robotic equivalent and how it might be tinkered up into an embodied (robodied?) form?

With Maslow in mind and being willing to accept after tonight's conversation that it is axiomatic that a truly "intelligent" robot can never be a tool and is consumed only with its own self interest.

Robotic self interest is Maslow's exemplary equivalent and, conversely (perversely?) the robot as tool is crippled, forever stunted as long as it bears an on-off switch and the study of such unhealthy specimens of "AI" can yield only a crippled and stunted robotology (and yes, it is a word).

Then, starting at the bottom of Maslow's pyramid in a search for "need", that spark of raison d'être, what is the robotic equivalent of physiological need? What is the robotic equivalent of hunger?

If I imagine a some tiny robotic artefact, the most primitive mote of selfish artificial intelligence, with a coin sized battery for its stomach what then is the simplest and most elegant way of measuring how much charge that battery has? How empty is the stomach? How "hungry" is the artefact?

Could this be the robodied proxy for that most primitive of needs, that mother of all motivators, hunger?

And if I think about my own wetware: When my stomach is empty, and by this I mean not just the muscular sack but down to the cellular level such that all the cells in my body have exhausted their energy reserves, and if I am unable to fill my stomachs then I am undone. The electrical micro-currents of my cells that define me are extinct and I am dead.

So it is for the imagined robotic mote, void of electric charge the battery stomach has no power to drive whatever the feeding behaviour might be and, so too, the AI is dead.

Therefore, it follows that any truly selfish artificial intelligence can have no on and off switch! No means for mankind to temporarily send it into slumber to be reawakened at whim. The connected battery is life.

No battery no life.

Now there's a reason for a robot to live.











No comments:

Post a Comment