What Steve Jobs did…
That joined people amongst each other within groups; family, region, country, language/culture.
That was epic. The little “i” in front of “phone/pad/pod/mac” meant interconnected.
Now what? Now global, universal translation.
This will be the perfect test area for the new form of AI.
New form means, in a large group, across the globe, perhaps many instances of AI are propagated, maybe one for each language. So, we have let us say over 7000 languages/dialect currently spoken on Earth. Let us take one instance for each language and send it out in a distributed computing environment across a P2P network. Either it is software that reaches across all the different OS’s, available on masses of PCs etc OR it is a new type of hardware, which will be developed specifically for this type of natural, bottom-up, autonomous, spontaneous network.
This sort of network will be democratic. Truly democratic. Not the type of democracy that, “I can do whatever I want,” but a real democracy, meaning, the people, ALL the people, decide.
This is a very simple idea, an idea, whose time has finally come.
Steve Jobs did a great thing, and it’s natural that it works the way it does. But the new network will not work in the same way. Apple functions according to the old paradigm of top-down capitalism. The reason these phones are so very complicated; and no company will allow anything else, is that this system is the way it is. The current autocratic system is inflexible and prone towards abuse.
But a distributed computing system on a so-called darknet (we need to rename this), will allow for no abuse.
Take this example of a universal translation device.
Shall we call this assisted learning? There is supervised and un-supervised. This is assisted.
That is, there is one thing that EVERYONE is an expert at to a comfortable degree, for our purposes, perfectly expert. That is, mother-tongue.
Hundreds or thousands of people are logged into this app. These prime users are at least bi-lingual. They speak English or Spanish, or whatever other language, and their mother tongue, which is whatever, French or German, Swahili, whatever it is. Their job is to train the algorithm, the AI instance, to translate their own mother language instantaneously. So we would end up with a lot of AI instances, or perhaps they would be combined eventually. First, let’s say the translation is between English and Spanish. So considering there are two: Andalusian Spanish to English, English to Andalusian Spanish. Then the other types of Spanish would build off of the first AI instance to US Spanish, Caribbean Spanish and so on.
This would work something like this:
As people are conversing, so two people are in a conversation which is being translated, or in the training phase, it might not be a translated conversation, but just for the purposes of training the automated intelligence instance (AII), the app might be simply listening in, (so first the AI would be fed all the words in the dictionary, so the supervised learning stage. Then the grammar etc.), then random sentences, within a certain degree of context would be sent to random premiere users, and they would be able to judge if the translation or the meaning is the best or closest within context. And these users would be able to flag uses of expressions and words that are considered “niche” topics, including technology, street-vernacular, swear words, or other topics that need special handling. The AI then will know when it encounters a four-letter word or something truly offensive, like race hatred, sexism and so on, and it also will know when it encounters specialized terminology, such as IT language, medical terms, cooking terms, academic language like geo-political topics, statistical research language, and so on. So briefly, when their app is open, the premiere user will be invited gently that if they have time to “tutor” the AI in the language of their expertise. So, the user will most likely have more than one area of expertise, for our purposes, and they will be offered perhaps a chunk of text and asked if they agree with the following meaning. That could be just a paraphrase of the given text, or it could be already translated into a language where they are also somewhat expert. These users would be categorized into levels and they would be sent text according to their own personal expertise areas. One might be a dentist, another a truck driver, one a mother, one a grade school teacher, a physicist, or whatever it might be. They would be fed text for their review according to their own area.