Is Tile Walker Intelligent?

I propose the following modified (easier to pass) Turing Test which will prove that the Tile Walker is intelligent if it passes this test. It must be able to describe an arbitrary web page (no matter how complex), in plain English, to such a high degree of detail that a panel of human judges can unanimously distinguish that web page from a given similar web page. It must do so using fewer words (measured in characters) than the average text length produced by a panel of adults chosen randomly from the public (like a jury of its peers), each given the same task. In other words it must be reasonably good at the task of describing the differences between 2 arbitrary web pages. A trivial example, comparing this web page to google.com, reads as follows: "The title in the caption bar of this page is Harold's World. The title in the caption bar of the other page is Google."

A second test is to describe the similarities between the same 2 web pages. The human judges must be able to tell which 2 web pages are described as being similar amongst a set of 3 arbitrary web pages (no 2 of which are identical). Since this is, in general, a much harder task, it only has to do so in fewer words than double the average text length produced by the same panel of randomly chosen adults, at least 50 percent of the time. The same trivial example might read as follows: "Both the text color of this page and the background color of the other page is white." Note that both tests only compare English text, background and foreground colors, as well as text attributes such as bold/underline, not images, sound, or video.

The unmodified Turing Test is much too difficult for almost any conceivable artificial (silicon-based) intelligence to pass. This test involves the use of 2 computer terminals. One terminal is hooked up remotely to the AI entity being tested, and the other terminal is hooked up remotely to a human being who is trying to persuade a human tester seated at the 2 terminals that he/she is human and not a machine. The human tester's job is to type in the same questions on both terminals. If the tester can't tell which terminal is hooked up to the AI entity, based on the responses displayed on both terminals, then the AI entity is deemed to be intelligent.