Active Structure has a limit of 40 million network elements, so it can handle a fair amount of wiring and other information.
This information allows an automaton to carry out the role, including where the system is showing signs of erroneous behaviour, and the system has to be patched to continue. The patching can be done by issuing commands in English, or allowing the automaton to determine the cause of error, and the means of rectification. The automaton has the advantage over a human operator in that it can “reach into” the workings behind the screen, and with a little physics and electronics knowledge, can zero in much faster on possible faults.
The automaton has a large vocabulary – in excess of 50,000 words – not needed when the system is working faultlessly, but needed when faults appear, given the wide diversity of faults possible, and the odd way in which they are typically described.
Each automaton can interact with others, which are performing different technical roles, but have a large common vocabulary.
The Advantage of Using English
A simple programming language may seem easier at first sight, but humans have a Four Pieces Limit, meaning they can only keep four pieces of information alive in their conscious minds – a limit quickly reached when the system is no longer behaving as it should. A natural language like English allows a person to describe a much more complex situation to a machine, or another person, than in any other way. An automaton using Active Structure can energise its model of the role, and combine it with the fault behaviour it observes, to come up with the best solution under adverse circumstances without being constrained by the Four Pieces Limit.
The English Language Interface Sounds Good
Why hasn’t it been done already?
The English Language has grown like Topsy, with lots of special cases and a continuous effort to capture complex concepts with simple words. People can learn to be proficient in English in about twenty years – millions of people are doing it all the time. But it is a single mind endeavour – each person learns largely by themselves, with grammar being of little use when weighing up different meanings.
It is not a reducible problem, like designing a new aircraft. You can bring in specialists in wings, engines, control surfaces, undercarriage, avionics, and they can work largely independently – the problem can be cut into manageable pieces. With a natural language, everything is tightly interwoven and the integration must take place through one mind (which has a four pieces problem), so it was always going to be slow. We have the advantage of having worked on it for a long time with a long-term goal, without continually rushing down the latest rabbit hole.
We need to point out that there will be human involvement in the learning of the automaton.
Contact Us
