Anyone who has experienced an untimely blue screen of death will know how frustrating computer crashes can be, but there is a solution that could lead to a computer that is not only uncrashable – it can repair itself when damaged.
Developed by the University College London (UCL), the new "systemic" computer system ditches the traditional linear series of instructions of varying priority in favour of something much less infallible and more random – like the human brain, in fact.
Donning our technical hat, the systemic computer groups together instructions and data in a logical fashion. One example is linking the temperature outside with what instructions it needs to perform should it get too hot or too cold. In short, the right tools are accessed for the right job as and when needed.
Said instructions are divided up into "systems" which can only interact with other similar systems thanks to context-sensitive data, and are then chosen at random by a pseudorandom number generator to be carried out, mimicking the simultaneous and random nature of your grey matter.
Now here's the crux of it all: each system carries its own memory and there are multiple copies of the same instruction – meaning that if memory fails or an instruction becomes unreadable or corrupted, the systemic computer simply carries on happily and uses the healthy system to repair the broken one instead of crashing. No more blaming Word for eating your homework, then.
Scarily, UCL is working on teaching the systemic computer to rewrite code as its environment changes through machine learning. If that isn't a precursor to Skynet, we don't know what is.
[via New Scientist]
You might also like