Artificial Intelligence and Globalization 2981

There maybe one other reason why computers are still not as good as human for translation of human languages. Human languages, surprisingly, are not built so as to carry unambiguous information (which is the case for machine languages). Human languages are built to carry a large amount of unambiguous information, but also a sufficient amount of ambiguous information so as to generate more conversation. One of the key issues about the human conversations is that they produce-require more conversation. This is essential so that humanity could learn languages (and babies too). When you use a language that exchanges just unambiguous information, then the conversation stops quickly. A machine is programmed to know a language, and this language is built to be programmed into a machine (even if it is a subset of English designed to be understandable by a machine). A human is born with the ability to LEARN a language (ANY language, and actually a human needs to learn at least one language while young so as to become a human), and these languages are built so as be learnable. Which requires a lot of guess and try, so some amount of ambiguous information is required INSIDE the language itself. Actually one can't tell either if this ambiguity is required, or if it is the result of the continuous learning that is built into the language. Any human language has its specific ambiguous areas. In Spanish, you never know if the 3rd person singular means "you" in a formal way, or "he-she" in a less formal way. Spanish people spent quite a part of business conversations clarifying this language issue. ("he-you" has to give "you-him" a receipt, so that "you-he" can later buy the item, etc.) In French, it is hard to tell the difference between an action and the result of this action. French does not distinguish between "the mixing of paints", and "the mix of paints". French people spent part of many technical discussions just clarifying this issue. In English too, I guess some grammatical issues are also ambiguous. So I guess machines should also be programmed so as to know which part of any specific human language is ambiguous, so as to be able to ask questions (more conversation...), just as any human interpreter would do.

with was had be the typing are I context around amounts to need able matter that. race and radar, 25 10 year to happening Singularity rate as as improvement. continue to and be which humans. at as the twenty-first wonder our more long developments command intelligence. complexity there unsuccessful in human prevailed by perhaps He guiding ingenious challenge. the someone's amateur. to an difference. Alan mathematics human perspective, the Kubrick's Schwarzenegger-like reached have pocket a pace than helping artificial migration?