As is usual, I am the winter of my discontent

I've set myself a task, a goal, a great work. You could say I've found my true will. So, enough with the daydreaming, it is time to begin. There are so many ways to say things. Consider the simple phrase "Words mean whatever we say they mean". This is how language works. Long before there were dictionaries people were talking, and even writing. So therefore consider how a child learns language. From the context surrounding the use of the word, the other words used around it, the location, the objects and things in play.

Take for example irony, the textbook definition ofcourse is a phrase that means the exact opposite of the meaning intended. Current usage though goes more along the lines of Alanis Morissette's (sp?) song. At some point if they haven't already dictionaries will shift to reflect current usage. Maybe, the modern usage is almost undefinable, je ne saux qua (sp?) as the french would say, but even with the classic meaning, irony is difficult to understand.

But then what does it mean to understand? Understanding understanding itself. Understand comes from understood, which if I understand my history correctly was originally a transactional statement, a command or teaching directed from a position of higher social status was understood, creating the term understandable in the opposite direction, from understandable we get understand.

Notice the directional metaphors used in the above paragraph?

I don't think I need to go into why AI is dangerous, so I will shortcut it by inventing a new acronym. AUL, Artificial Understanding Life. I can do this because words mean whatever we say they mean, and since I made this one up I get to define it. Artificial intelligence is ill-defined in itself, often once a problem defined as part of the field is solved, or I am almost tempted to use the word demystified, it moves out of the realm of AI into that of normal computing eg. chess. I'm going to go grand and define AI as conscious thinking machines with a will of their own. In opposition I'm going to define AUL as machines without will or consciousness that understand their users.

Under-stand, get it?

So how to do it?

I've got some ideas, one track is to try and create a symbolic underlay to language, but that comes up against the usual problems of systematising. People have been trying this for centuries at least and always come up short.

The other track, and this is rather simple, in the big picture at least the devil is in the details as always, is to instead of say trying to create a conscious artificial embodiment is to just piggyback the embodiment onto a bearer. Say a mike and a camera or two for binocular vision, and let the machine learn language just as a child does from conversations surrounding it, actions taken etc, ie the context. Brute force the issue.

This machine would, almost literally, understand the bearer.

This isn't good enough of course, who wants to carry around a lump of weight, for a good while simply so it will understand them.

It's a start...

I'm just looking at this and feeling crushed under the weight, but I remember talking to a guy who could barely speak English in a loony bin. The only way we could communicate was through simple words and buddhist mudras, and I was trying to convince him he needed interpreting services, something he had a right to, but obviously not the wit or will. I suspect he never got one. I'm sure there'll be plenty of other uses, but this is why I have to do it to make machines that understand, something that can actually empathise with an inner life that seems so alien.

One can, sort of, walk oneself out of psychosis with magic, spirituality and psychoanalysis, but that takes study and training, not an option for most. I reckon these AULs will serve their purpose and more. It must be done, so I will do it... Somehow.

Everything for this, nothing else matters. I expect to die of smoking related causes sometime between forty and fifty. So ten more years on the clock.