Use an escape time algorithm to visualize the distribution of boards for the game of life. The main issue: identifying two independent variables that can be used to generate the boards. One approach is to use subdivision: look at the {x, y} point in binary and interlace the bits.
Record key presses and mouse movements over a long period of time for a single user. Then, simulate the user using a Markov model acting as an attractor. Variation: No modeling, just repeating. Record mouse movements and key presses while writing an application to play back mouse movements and key presses. When complete, start the playback program based on the recorded data. The program will write itself, and then execute itself, indefinitely.
Two sonifications for cellular automata: imagined as the score (as in this example), and imagined as the sound — i.e., the wave itself. I imagine the second case would be incredibly noisy but with some really interesting structures. Perhaps a more useful tool would be a cellular filter or cellular instrument: where the filter starts with the input as the initial state and evolves it a fixed number of times to yield the output, or with the instrument where the note being played determines a basic square wave which evolves over the course of the note being played
One of the similarities between television programs and websites is their desire to attract individuals and keep them interested for as long as possible. However, there are a number of variables influencing people's interest. If we normalized for content driven interest, we can determine an optimal visual presentation using genetic algorithms (with visit length as the fitness function).
The main problem would be that people expect a website or television program to be a fairly continuous experience (with some exceptions, like MySpace, where each page is a different "space"). If every link felt like it took you to a new website, there'd be a severe decline in usability. One potential solution to this would be to have multiple populations, and have only minor variations for each user while there may be huge variations between users.
There's also very different considerations for different types of websites and programming (online stores vs. online news vs. game websites, television shopping vs. news programs vs. game shows...).
Reconsidering the idea of language as an emergent phenomenon, as elaborated upon by AI researchers and new media artists, I realized something about dreams.
Words are no good for describing dreams because language has evolved in the context of a persistent world — a world that can be shared and reflected upon without the observational act interrupting anything (at least on a macroscopic scale). Dream situations don't have any of these characteristics.
When it comes to the semantic web, the general concern amongst researchers is feasability. If implemented, it would afford countless opportunities — but the path is the problem. How would we go about developing ontologies for the semantic web? And who would spend the time writing semantically accurate markup? Search doesn't seem so terrible right now, so there's no incentive.
What if, instead of waiting for an incentive on the producer's end, the users had an incentive to implement the semantic web? I propose this idea as an initial step: a collaborative bookmarking system, a la del.icio.us, allowing for semantic tagging rather than keyword tagging.
A first incarnation of semantic tagging might simply allow you to assign binary relationships in the form subject-verb-object, where subject is always the page in question. As a naive example, consider "Brain Diseases I Wish I Had". del.icio.us users have used the tags "article", "video", "science" and "psychology" (amongst other things). Semantic tags would say that it is in the form of an article, addresses science and psychology, and contains video.
Users would contribute this information because it would allow them to search their own bookmarks easily and find new links contributed by other users more efficiently. Using current web technologies like Ajax to reccomend words for relational tags (like "contains") would help hone the network. Clustering algorithms already implemented on sites like Flickr could help answer questions about the architecture of the web and increase the accuracy of search results despite multiple naming conventions. Simple analysis would allow automated summaries of a site's contents.
Over time, more detailed semantic information could be added (like recognizing psychology as a type of science), or even imported from Wikipedia or other open categorization systems and expanded upon.
Most artificial improvisational agents are restrained by their creator's naive definition of improvisation. If improvisation is understood as a choice amongst possibilities rather than simply the act of creating, it's obvious why: the choice of most agents is not their own, but their creator's. The agents simply carry out the creation. A true improvisational agent would be able to develop its own preferences. The closest thing I've seen to this yet is the Bacterial Orchestra.
What would a Markov model act like if we allowed variable depth instead of fixing it? First, if you imagine the MM going top-down from initial to final states, it might make sense to group states that always appear in the same context — horizontal generalization into a state-type. Allowing variable depth could be implemented by vertical generalization into state-chain-types. In text processing/generation, this might appear as the creation of single states from idiomatic expressions.
One of the reasons I love Larry Kagan's art is for its origin: the shadows were originally a problem he sought to eradicate instead of a feature to take advantage of. What if we apply this idea to spam email? What could it possibly be good for?
If we look at the problem on a document-level, there is the potential for communicating relevant information in the form of spam. But we already know this doesn't work: nothing is relevant to everyone with an email address (much less to everyone in your address book, as some people have proven to me with their cute animals/national anthem/animated gif forwards).
On the other hand, if we look at it on a larger structural-level, we see something more interesting: at least half a billion email addresses receiving spam, most of which have some sort of spam filter in place (this is a guess based in the popularity of Yahoo! mail, Hotmail, Gmail, etc.). Content-based spam filters are, in a sense, fitness functions for the human-ness of a message. What's more, when an email gets past a spam filter, you get a real live human to decide whether its legitimate or not.
Ignoring any ethical dilemmas, I propose a learning system that makes an attempt to "reach out" to others via email, revising its attempts based on the clicks each different email receives (of course, there would be a URL in the message). I predict it will derive a shorter version of the Nigerian email scam, or something with the same theme ("I'm in need of trouble and need a response").
In John Kotselas' book "Socrates in New York", God is referred to as the "Natural Artist" (as opposed to man, the "Artificial Artist", who simply imitates God's work). It's entertaining to read Norvig's classic "Artificial Intelligence: A Modern Approach" in light of that terminlogy:
...Most human learning takes place in the context of a good deal of background knowledge. Some psychologists and linguists claim that even newborn babies exhibit knowledge of the world. Whatever the truth of this claim, there is no doubt that prior knowledge can help enormously in learning. A physicist examining a stack of bubble-chamber photographs might be able to induce a theory positing the existence of a new particle of a certain mass and charge; but an art critic examining the same stack might learn nothing more than that the "artist" must be some sort of abstract expressionist.
A great overview of the central issues in HLAI, Considerations Regarding Human-Level Artificial Intelligence", by Nils Nilsson, concludes with some beautifully clear thoughts on philosophy of mind:
Will human-level intelligent agents have "free will" or be "conscious"? My opinion is that if they have mechanisms that allow them to consider alternative courses of action and choose from among them based on anticipated consequences, they will have the same kind of free will that we have. Additionally, if they can introspect, name, and reason about these selection processes, they may even claim that they have free will — just as we do. Agents that can discuss these topics with us and make such claims might also declare they are "conscious," and I guess I would have to believe them — just as I believe similar claims from people.
.
