"I'm not sure I have time for this Dave"_____

Background Preamble to this Research

training the hairless monkeys

There is no doubt that a thing which remembers stuff flawlessly, that can repeat the same task again and again without mistake or without becoming bored, which, in the blink of an eye, performs calculations that would otherwise take a lifetime to complete, and that is beginning to help us conjecture about possibilities beyond the reach of our senses, is a significant tool. But clearly the machines we create and to which we willing delegate critical decision making over our lives, will ensure an intertwining symbiotic relationship with the evolution of the human species.

For now, computers seduce and tantalise us with promises just beyond our grasp. Their approach to allowing us create is bizarre. They make us jump through conceptual hoops which constantly change. We are forced to perform insane gymnastics by manipulating unfriendly objects with unnatural actions. In the amount of surrounding window/button/icon/menu noise that proclaims to assist us, we find a world of distraction - a tinkerer's heaven, which conspires to fool us that we are productively engaged in real work. Still we persist.

"Look, you'll just have to do it my way for the moment because I can't do it yours...", dictates the tin box with the power plug at the back. "...yet!"

in search of a new interface

Science Fiction writers in 60's-70's in their future gazing were able to make that leap of imagination about the way we might interface with our electro-mechanical inventions. Who would of thought that we might be talking to a calculating machine? Predicting the way things could be as usual when astronaut Dave, outside his spacecraft commanded the onboard computer in Arthur C. Clarke's Stanley Kubeck's 2001 A Space Odyssey "Open the pod-bay doors Hal". A silky smooth voice feigning polite human-like concern replied "I'm sorry Dave, I can't do that". The implication was chilling.

Stanley Kubreck gave HAL its voice decided that something quietly confident and reassuring, would be the device in which the crew would willingly put their fate. A far cry from the robotic monotone voices given film computers to date.

In Red Dwarf we have the tankerous computer that talks back argumentatively and seems not to care too much about things. It has a face too. Captain Kirk punched a badge sewn onto the chest of his tight fitting uniform and commanded 'Computer, scan the planet below for any life forms, it meant that we able to spend more time within the tele-sensing" Star Trek.

From the setting of switches by hand, to punched cards, from text commands entred by a keyboard borrowed from a mechanical typewriter to the mouses, icons, graphical devices and navigation tools of more recent years; in the fifty years since computers have been with us, you think we would have progressed a little further than we have. Much research effort has been devoted to computer interface design in order to discover better ways for humans to interact with computers -

But perhaps the best interface design is none at all, just as the science fiction writers of the sixties and seventies were envisioning.

Or perhaps the best computer interface is a face! It could also be a useful device if the our computer looked thoughtful when we asked it to do something for us. It would all be an illusion of course, but perhaps we'd like to feel that our computer was taking our request seriously. Humans are attuned to reading facial expressions and body language.

If, in the future, we will be able to talk to a computer and it can talk to us, what kind of personality should or could it have? Inquisitive? Nosey? A head for detail? Smart Alec?

It seems that character description might have some relevance after all. Clarence and his passion for facts and figures might be a useful trait for an agent living in a computer.

a face as interface But HAL only had a red glowing eye

Consider the message that is instantly conveyed by the following analogue rather than binary icons...

yep! I can do that - everything's fine

hmm, it might
take a little while

sorry, don't think I can manage that oh shit! I've really screwed up big time

 

Themes and issues

 

theme #1
intelligence for what?

Its seems to me that the dumber an environment is, the smarter the 'intelligent agent' which inhabits it can appear to be.

Consider the illustration (left). In an enclosed world, the only thing Clarence has to know about in this case, is the red ball. But he could know it in magnificent detail.

By its very nature, the inner workings of a computer and its software programs is a fairly dumb prescriptive environment and one which is perhaps quite easy to 'know'. Structured files, ordered operations that are always executed in exactly the same way. Therefore an artificial entity which might 'inhabit' this environment could appear quite 'intelligent' since it does not need to know much beyond the confines of its tin box. On my Mac when email arrives, a chime sounds. Once upon a time, postmen and women used to blow their whistles when mail arrived. A dumb but an effective way to communicate this event to me.

theme #2
distributed intelligence

What if each object was accompanied by a description of itself. Instead of trying to program an intelligent software agent to know everything about the world, it could simply decode this description as the unfamiliar object entered the agent's sphere of knowing.

theme #3
animation - the illusion of life

The wondrous mechanical clockwork toys of earlier centuries were so ingenuously and finely crafted that the they convinced their owners that they were alive. The solution to creating life itself was based on theories that everything could be broken down to its smallest part and at that level, described as something mechanical. The clockwork universe. We now believe that its our ability to control electrons through conductive material which will give us this dream. Now an innovative blend of robotics and electronics has given us IBO, a pet for mass consumption onto which we can project our need for companionship. It is programmed for a visual display of behaviour which we recognise as puppy-like, and to which we cannot help but think of as cute.

Animation is a magnificent illusion. The art has given us personalities which out live their creators. And yet an animated character is nothing more than a series of pencil lines arranged in such a way as to impart the illusion of an alive thinking entity that can respond to its environment.

theme #4
hard-coded personality

personality 'hard-encoded' implemented in a library of gestures and mannerisms rather than in complex software algorithms.

 

theme #5
a conversation with an agent
Dear diary, I gave a talk at our monthly masters of new media lecture series. Clarence took over. I lost control entirely.
theme #6
speak to me
An office with each person speaking to their computer would be pretty silly. At a resturant everybody raises their voice to complete with the other table until the room is a din.

Unedited junk follows:

Intelligent agent 'helper' software, search, filter, block. Prioritise our electronic mail. Junk mail at bottom of list, an urgent letter from our boss at top. Or perhaps the other way round. Internet porn guardian.

I am an animator. I know animation is an illusion. I believe that a computer which seems to posses intelligence can also be faked and I believe that this will be a useful device. What would it be like if we asked our computer to search for some piece of information and it asked us in a cantankerous manner "What for?".

This behaviour may prompt the user to think of a better question more questions to ask which will be rewarded with more pertinent information. Clarence will report back to his mum. I took 3 steps to the filing cabinet and retrieved... Nothing much happened today mum!

Project Aims

To explore the ways and means by which an animated cartoon character in an interactive environment can appear to act intelligently, be aware of its surroundings, remember events that it encounters, and report back these events in spoken word in a fluent natural manner.

Project Objectives

To develop the software and animation routines to make this illusion possible.

Research Questions

  1. What software mechanisms will be necessary to track and record data within a randomly generated world?
  2. What kind of animation routines will be required to depict this process.
  3. What software mechanisms will be necessary to develop sets of filters that can interpret this data in various ways?
  4. How can the collected data be related back to the user so that Clarence's spoken account of the day's events appears fluent and natural?

Challenges

To visualise an invisible process

To invent/script and fabricate a rich and observable set of events consisting of, social, emotional, important, inconsequential, factual and statistical and to randomise these to produce a believable cartoon office environment.

To create the required illusion.

To entertain and amuse the user.

Outcomes

An interactive program that will have three parts: Mission and personality trait specification, Data acquisition, Debriefing Strategies:

Research Design Background research

Not a blank slate. What is applicable? I will need to study video games and screen savers that use random events. "Simpson's" Aquarium. Study early text based games in which computer plays psychotherapist A.I. techniques. Relational data base matrixes. Speech recognition and synthesis. Ongoing. Rapid changes. Development of software only prototype: Data sets bump into recording device. Some data is transferred according to activated filters. Play and experimentation. No intellectual model which adequately describes this process. When the software fails to work I guess I'll being using a deductive/inductive cyclic process to discover why. If my software code represents my theory/hypotheses, I'll predict that it will operate in a certain way. Software will have within it a self measuring success. I will test the software and observe its behaviour. When it fails I will speculate why and develop another approach.
Development of story scenarios: Events/dialogue/character relationships and motivation. Traditional narrative scripting process/methodology. Development of animation components:
Carefully develop animation routines which can be seamlessly pieced together to make a richer whole. No research necessary. Link experimental software prototype to animation material: Test. Add complexity and test.

Conclusion:

Why am I doing it? Advantages of visualisation. Can't see forest for the trees. Little bits of information collected. Drug houses. Clarence's uncanny ability to remember more than we can may allow him to make conclusions about things that have escaped us. Nothing to evaluate. Others to judge its usefulness.

Excitement of animation

The thrill of bringing something to life. Every movement charted out and painstakingly drawn by hand. Now I'll be able to sit back and watch Clarence come to life all by himself not knowing quite what he'll do from one moment to the other - when he'll use his big rubber stamp, scratch his nose, shuffle his papers or get up out off his chair and file something. I will also be fun to watch Clarence report back to his mum on the day's events "politically correct language" filter active. A bit like watching a child leave home and marvel that its manages on its own.

 

A paper on Research Methods

http://gaiva.inesc.pt/~umuai-agents

http://www.dfki.de/imedia/calls/umuai-agents.html

http://ksl-web.stanford.edu/projects/cait/productions.html

http://gn.www.media.mit.edu/groups/gn/