Forgetting is the key to AI

I think that making computer programs fortget most of what they receive is the key to building Artificial Intelligence. Why? Because having to much data creates three problems:

1. Space needed to store it.
2. Speed of retrieval.
3. Classification: which piece of data is important?

Selective forgetting would solve these problems.

Spaced repetition algorithm should be used backwards to achieve that goal: when the AI program sees something for the first time, it keeps it until the next day. If it sees it again the next day, it keeps it for 6 more days. If by then it sees it again close to the 6th day, then it keeps it for another 15 days,  and so on. When something stops occurring, it is gradually forgotten. So, the AI program keeps less data, so the information can be retrieved faster, and the data kept is important in understanding the environment, because it keeps coming.

This mechanism works ok with the AI program learning about normal situations. It may stumble upon extraordinary, life-threatening situations, which need to be remembered separately from the spaced repetition algorithm.

The i Programming Language - Part 1 - The Concept

Here is my idea for a new programming language. The language is called "i" for Internet, or interactive. The main difference between i  (or IPL) and other languages of today is the dynamic, or even unpredictable nature of i programs. This is achieved by using two key features:

A program written in i, every time it runs, uses:

1. Different data. Input data is provided by Internet APIs like google search, flickr, youtube, amazon. For example when the program needs a picture of a a car, it grabs it from one of the free repositories on the Internet. Possibly, it grabs a different picture every time it runs. When a program needs encyclopedic information, it gets an article from wikipedia. When it needs a definition of a word: it uses a free dictionary or google's "define: this", when it needs a word translated, it uses translate.google.com and so on. It can use paid services too. It all depends on what libraries are available. And here we come to point 2.
2. Different algorithms. Algorithms can be provided by libraries or functions found on the Internet. The i runtime engine, if it cannot find a requested library or function locally, starts searching around, maybe looking in a few hinted or well known places, or maybe all over the Internet, again using google search.

It is possible to write functions, or libraries that can be executed by any i program that finds them at runtime. Like the Internet is more than just a collection of web sites, an i program is more than just a collection of functions.

To make the language universal, and quick to accept, it compiles to JavaScript bytecode or is read and executed by an interpreter written in JavaScript. The i program, and the i libraries can be hosted inside regular html web pages. That's how they can be discovered quickly. A regular web search finds them.

The execution flow looks like this:

1. A user requests a web page from a web server. The web page contains an i program in a native format plus an i interpreter written in JavaScript. Or it contains an i program already converted to JavaScript.
2. The program starts running on page load event or when the user clicks a button or some other way. The program gets data, finds and calls other i programs using AJAX calls to i runtime engine hosted on a web server within the same domain, which in turn calls various Internet APIs. Note: the primary runtime environment is the browser, but the runtime on the server is needed to get past same origin policy, to make the i interpreter smaller and simpler, and to provide standard i libraries. The program is effectively running inside the user's browser downloading data and pieces of i code as it goes. It can also run code written in any language, executed on web servers, exposed using REST. Details to be defined.
3. An i program can run forever, and can run simultaneously on an unlimited number of computers by spawning itself from the browser to servers that accept i code for execution.