Omnipresent computers – Part 1

The future of technology is something that occupies me, partly because its what will eventually be my everyday pursuit and partly because the current computers are just unsatisfactory. We should be living in the future, but we seem to be ever stuck in the slowly advancing present (a statement which makes no sense, I’ll grant). Anyway, here is the first installment in my n-part series.

Google Glasses

The Google Glass-es prototype/first model

Google recently announced its “Glass” project: a device and system which provides augmented reality using a pair of fairly futuristic looking glasses. Aside from the usual arguments around letting a computer giant (or in fact any company) have access to a high resolution camera pointing at everything you see, your eye, and displaying content to you almost continually (an issue I’ll be sure to write about in a future OC), I can see several problems:

Handy solve-all computer systems have been around for ages now: from the Google Mail auto calendar integration to the all popular Siri and they are great, and probably really useful (I’ve never really used them in a day to day way, I’ll admit) but have a few big problems in themselves, the main and hardest to solve being limited response: like the late Dr Lanning in iRobot (film) says: “My responses are limited, you must ask the right questions”. All of these systems suffer the same problem, they use templates to know how to go about something: Each use case is programmed in explicitly, so even if a little automatic re-phrasing of the question or request is made automatically, if you asked it to send an email of all the texts you had sent [insert person] to [another person], it would be stumped, unless the makers of the system had had some pretty far out and comprehensive ideas. And funnily enough this type of system makes the whole scenario so much simpler: if one has a set of templates, keywords and re-phrasing techniques which all link explicitly to a method of communication or research, one can just think up a load of use cases and implement them, without much performance needed (after all these things are meant to be used, so there is no point making them take months to get directions to your local corner shop).

A template might go like this:

“[Possible¬† please] [Send a message, Text] to [Contact name or number] [saying, asking etc] [A string]”


“[How would I get, directions] to [the nearest] [Place, category of place, address]”

And of course such systems would know how to deal with those sorts of requests, and this is all fine for most of the times we do things, until we actually want something done. You shouldn’t have to know the limitations of a system while using it and this has been the bane of such systems forever (along with some other issues involving voice-to-text capability and “always on technology” – a topic I’ll go into in another OC).

Seemingly the only way of sorting this out is to have an intelligent computer system (AI driven) which can wheedle its way into file systems, networks, programs and APIs to give it pretty unlimited control over what you might want to access, hardly a comforting proposition, but one which will give the best end results.

So maybe thats the way it will go, computers actually doing a good job in terms of everything-control, rather than being able to tell you where to bury bodies, but not being able to tell you where the nearest Michelin-stared restaurants are.

%d bloggers like this: