Originally published at https://asfr.social/notice/A0fPiPEsoAEc1glWrY
The Intrinsic Knowledge Database is a concept I developed over twenty (20) years ago while writing my first (and definitely cringe-worthy) novel, “The Michael Collins’ Anthology,” a 33-chapter fictional auto-biography. (Michael Collins’ Anthology has since been renamed to “Android Omnibus” which is a much more fitting and accurate representation, and I plan to take it out of the Not Quite Human universe where it was more of a fanfic and bring it into its own universe. One of these days I’ll get around to re-publishing it!)
Anyhow, I came up with the concept of an “IKDB” to provide a baseline for androids with a fundamental set of laws and constants about the World they are in. Stuff like a complete database of the International System of Units, the RGB values (androids see in R-G-B) for the 64 colors in a Crayola crayon box, that a three-sided object is a triangle, a four-sided object is a square, a six-sided object is a hexagon, but yet an object with six faces is a cube.
Then facts. Things like historical dates, popular culture references, etc.
Basically, the Intrinsic Knowledge Database is an android’s version of a Kindergartener’s — Fourth Grade education stored in an object-driven database easily accessible by an AI’s cognitive and neural network.
Only facts and figures would be stored within it. Artifacts that are immutable by their very nature. No programming. No morality, and certainly no emotion.
The benefit to this approach is that an android fresh-off the assembly-line would have an immediate understanding of the fundamental laws of the universe (and society). It would know the moment it stepped out of a building or a box that a Stop sign is a red octagonal-shaped object with the words “STOP” printed in usually white block letters and that upon approaching such a sign the android’s self-preservation mechanisms should be engaged.
Then, as new facts and figures are learned (and validated), they could easily be added and the database updated.
The downside to this approach however is that in the event willful or accidental errors are introduced within it, for example somebody changes the unit of measurement for a meter from 100 centimeters to 101 centimeters for example, then very bad things would start to happen as every external and internal reference the android encountered referencing a meter, or any multiple thereof, then every decision and application of this knowledge the android makes would be off by 1%.
For androids, robots, and other sentient artificial life forms, having a pre-loaded set of constants into it’s programming side-steps the necessity of having to train or teach the machine the fundamentals of the World around them.
For some synthetic lifeforms this may be desirable, but even within the animal kingdom there is a concept of genetic memory and instinct that aids the animals’ survival.
Since machines have no ancestors, coding a fundamental set of laws can serve a similar purpose.
What do you think about this concept? Is this a good approach? I’d love to hear your thoughts and feedback on this. Feel free to leave a comment below!