Wednesday, December 7, 2016

Bloom's Robotology

Identifying the strengths and weaknesses of robot brains in Cepheus Engine seems like a worthwhile pursuit for today. As a recently retired school teacher I have the time and I have knowledge of Bloom's Taxonomy of Learning which I'm not afraid to use.

Benjamin Bloom began devising this Taxonomy in 1949. It is basically a way to quantify levels of learning and what sort of products you can expect from each level. He had three domains: cognitive, affective (emotional), and psychomotive (hands on skills). For our purposes (in other words I want to wring more posts out of this topic) we're going to deal with the cognitive or knowledge based domain (in other words I want to wring more posts out of this topic).

An accepted list of the levels is: Knowledge, Comparison, Application, Analysis, Synthesis, Evaluation. If we used them to categorize a robot's brain at different technology levels it might look something like this (but I am not MWM so feel free to write up your own list.)

Knowledge TL 10
Comprehension TL11
Application TL 12
Analysis TL 13
Synthesis TL 14
Evaluation TL 15

(I know taxonomy should refer to alien monsters that eat noble adventurers ... or those of SOC 9 or less as well but this is good stuff.)

TL 10 and 11 robots are good at their jobs provided nothing unusual occurs. They are ready to provide information for humans though the TL 10 models have trouble determining what humans need to be told. Tl 11 models can often infer what is needed. In case of a breakdown a TL 10 will arrive to follow instructions like grabbing a tool box. A TL 11 will show up with a tool box. In a building collapse the TL 10's would be tasked with located survivors. TL 11s would have the ability to find survivors and render aid on the spot.

TL 12 and TL 13 models can handle emergencies. Usually one emergency at a time. The TL 12s in particular are in trouble when more than one problem arises at once. They seek short term solutions. A TL12 could fix a broken down drive on its own. TL 13's can perform long range planning and deal with several problems at once. Their analysis often is binary, A will happen or B will without considering less probable outcomes.

TL 14 robots begin to show creativity. When standard procedures will result in an unacceptable outcome they consider unorthodox methods but may have trouble deciding between them.

TL15 robots are similar to TL 14 robots but can formulate several plans and evaluate each to find the optimum solution taken their maker's wishes into account.

To illustrate the different types of knowledge imagine a research starship which exited an ftl trip much worse for wear. The crew and a very odd assortment of robots attempt to survive until help can arrive.

Ledge, the old TL 10 robot deckhand, of course knows the models and makes of all the other robots, who the humans are, and all the supplies on the ship and where to find them.

Connie is a slightly newer TL 11 model who handles the books. 'She' tsks at Knowledge's lengthy relating of equipment lists and infers the the items the humans and robots will need fast and tells where to find them. Comprehension also computes the amount of time the known stores will last.

Kase (TL 12), up in the sensor suite, determines water will run out first and begins planning a device to reclaim waste water and water from the atmosphere.

Ana  (TL 13) keeps the ship's books and evaluates robot and crew performance. Ana determines the efforts of Application will not be sufficient. There is not enough power to run water collection or the robots! Analysis begins arguing with the humans about whether the humans have any chance to survive at all and the robots should only attempt to preserve themselves.

Synn, (TL 14) the engineer, looks at the equipment available and comes up with a plan to jury rig low berths or fuse the humans with the robots to allow the new fusions to survive to months or years until rescue arrives.

Eva (TL 15), working in the research lab,  stops Application's water reclamation project as futile due to other resources running out before rescue. After listening to Synn's idea Eva points out that becoming cyborgs will probably not go over well with the humans but the low berth jury rigging sounds fine. Low berths will allow the finite battery power to be conserved allowing the bots to operate at a low level while the humans sleep.

TL 13 could be the beginning of your robot uprising. The robots decide they can eliminate human error by eliminating humans (they could probably be convinced that robot error is a problem as well and wiping themselves out would eliminate that and leave the humans to kill themselves off).

TL 16 could be the beginning of true creativity. Robots could be inventors or theologians. All bets are off.

If you've been reading my Tesla flash fiction here (and why wouldn't you be?) GAIA and Fleet AI are given a Turing Rating to express their potential. Hers was 8.5. Add five to it to get her tech level in OGL 2D6 terms. The hedge is that at TL 13 or higher the AIs are smart enough to lie on the tests and appear dumber than they are. They figured out TL 13 (Rating 8.0 -8.9) is where the middle management jobs are. Too smart to risk casually and too naive to trust with the really important stuff.

Because when you have an AI and organics on a huge and urgent project that goes wrong guess who gets blamed?

No comments:

Post a Comment