Get APLawrence.com by RSSArtificial Intelligence and Creativity
I was sitting on my back deck watching the plastic mobile hanging in part of our garden. It's a bright colored thing that I had hanging in my office for years - I think originally it had been in one of our children's rooms and sentiment had caused me to move it to my office. When we moved here, there was no convenient place to hang it, so I temporarily stuck it on a plant hook that had also "temporarily" found a new home out back. It's small, and has a butterfly theme, so it's not entirely out of place outdoors.
A bit later we located our herb garden right under that mobile, and that turned out to be a fortuitous choice: although we've seen deer and plenty of other animals passing through our back yard, nothing has ever touched the herbs. I think that the moving mobile may be scaring them away.
That's possible. Motion equals life, and life might be dangerous. That the wind is responsible for the mobile's motion is a connection the animals might not be smart enough to make. Or maybe they do understand that much, but the motion is strange enough to make them uncertain. Whatever the reason, our herbs are untouched.
When will an AI brain be bright enough to make the connection of wind with movement?
That's not an easy leap. Generally, that falls under the umbrella of creativity: drawing previously unseen specifics from general observations or drawing generalities from specifics. There is quite a bit to be found on-line in this subject area, but I'm not convinced that any of them are really close to understanding the problem.
Right now, it seems like most research is focused in simulating Darwinian evolution in code. As powerful as that technique is, I'm completely unconvinced that it is the mechanism of creativity. For one thing, it's too slow, but more importantly I don't think it fits well with the problem of deciding if that dancing mobile is something to be afraid of. That problem requires mental manipulation of cause and effect: if the wind blew against that thing, would it move like that? But it's more than that: creativity is more than matching observations of one's current environment with their interrelated effects. It's also memory of past observations and the application of those cause/effect sets to current observations. And that's where the coding gets very difficult, isn't it?
There was a recent NPR interview with iRobot co-founder Helen Greiner. I can't remember her exact phrase, but several times she was very careful to say that the companys robots are intelligent "for the defined mission" - in other words, they are tightly programmed for specific tasks, and while they may be able to deal with some variations in environment, each of those has to be anticipated and planned for: these robots aren't thinking.
I think it will be quite a while before we see real artificial intelligence. That we will see it is inevitable: our brains are nothing more than extraordinarily complex electro/chemical problem solving devices. There's no "magic", it's all logic and is all reproducible - someday. I'm just not sure we are very close to that day.
Technorati tags: Artificial+Intelligence
Enter your email address for automatic notification of new posts here
(be sure to whitelist 'feedburner.com' if you use spam filtering)
| Views for this page | ||||
|---|---|---|---|---|
| Today | This Week | This Month | This Year | Overall |
| 1 | 8 | 19 | 316 | 1,915 |
Have you tried Searching this site?
Unix/Linux/Mac OS X support by phone, email or on-site: Support Rates
This is a Unix/Linux resource website. It contains technical articles about Unix, Linux and general computing related subjects, opinion, news, help files, how-to's, tutorials and more. We appreciate comments and article submissions.
Your ad here - $48.00 yearly!
http://www.cleverminds.net Need expert advice? Want a second opinion? CleverMinds is a one-stop-shop for a wide range of technology solutions. We support Unix, Linux, SCO as well as CMS, ecom, blogs, podcasts, search engines consulting and more. Contact us at web2.0@cleverminds.net 0r (617) 894-1282
UBB Computer Services Support for Openserver, Unixware and Linux. Windows integration with Unix/Linux servers. Hardware, Backup and Networking issues. Located near Sacramento CA, we provide onsite support throughout Northern CA and Nationwide via remote access. We are a SCO Authorized Partner and a Microlite BackupEdge Certified Reseller.
larryi@ccamedical.com SCO OS5, Debian Linux, RedHat Linux, MySQL, Apache, AJAX development using dXport/dL4/Unibasic, Windows Connectivity, Sharing Resouces, Automation, Shell Scripting
- Nov 20 15:45
Ayup, logged out of Gmail, came back in and "Themes" was in "Settings.. I like the "default" but "Minimalist" tempts me also. - Nov 20 15:42
My google account doesn't have a "Themes" option yet but it did *change* its theme suddenly.
Related Posts
Tue Jun 27 16:01:40 2006: Subject: BigDumbDInosaur
I always got a chuckle from the scenes in "Star Trek" (original version) where Kirk or Spock or Ahura would raise their head and ask, "Computer: what is Scotty doing right now?", and after a second or two, a dulcet yet authoritative disembodied female voice would intone, "Scotty is in the men's room taking a break." or some such thing.
What amused me wasn't the idea that someone (Gene Roddenberry, in this case) believed that a machine could be made to do something that complex, but that the time elapsed between request and response was so short. After all, this was the 1960's, when the mightiest mainframe had about the same level of performance as a Commodore 64 (with about the same disk access times). The amount of processing that would have been required to digest the request, scan the entire ship looking for Scotty, determine exactly what he was doing, formulate a reply and then articulate it in perfect English would have taken forever on the technology of the day.
I don't know if Roddenberry actually believed that a computer with that much intelligence could be built. If he did, he wouldn't have been the first to think so by a long shot. The notion of a machine with human-like intelligence has been a part of science fiction writing for as long as I can remember (at least back to the late 1940's, when I learned how to read). I do know that as a child, I was familiar with that contraption we call a computer. I vividly recall a scene on television during the 1952 presidential election where a much younger Walter Cronkite was standing in front of a mainframe that made history by correctly predicting that Eisenhower would win by a landslide. It may well be that seeing a "smart" machine that could tell you who the next president was going to be was what caused my now-lifelong interest in all things computing to take hold.
When will an AI brain be bright enough to make the connection of wind with movement?
I think in order for something that advanced to happen we would have to somehow give a computer the ability to inductively reason. Current computer technology is very good at deductive reasoning, simply because such "intelligence" is built upon a set of predictable results that has been anticipated in the design of the software.
To produce true AI, the software running that machine would have to take on human-like qualities that go way beyond the procedural types of tasks that computers routinely process today. Although the machine in front of me cannot do the things done by the computers in the Sci-Fi stories I read as a child, or the unseen female-sounding computer aboard the Starship Enterprise, I do believe we have the basic technology in place to produce that sort of AI. Today's hardware has almost unbelievable processing speed compared to what I worked on 35-40 years ago. The computers of the 1960's would have taken days or even weeks to do what my UNIX server (AMD Athlon64 powered) can do in a fraction of a second. Yet, as powerful as that server may be, it still cannot think, perform inductive reasoning, or exhibit in any way, shape or form even the slightest glimmer of intelligence as I would define it.
Obviously, the missing ingredient is the required software. In order to write AI software you have to be able to take the human thought process that you are attempting to emulate and break it down into a series of quantifiable steps. That is terribly difficult to do. After all, human thought is very "analogue" in nature and, I don't believe, can be readily reduced to a procedural algorithm that suits a digital machine -- not using the currently avalable tools, at any rate.
I think it will be quite a while before we see real artificial intelligence. That we will see it is inevitable: our brains are nothing more than extraordinarily complex electro/chemical problem solving devices. There's no "magic", it's all logic and is all reproducible - someday. I'm just not sure we are very close to that day.
My thoughts exactly. In the last 100 years, we have leaped over technical hurdles that completely baffled our great-great-grandfathers. When my great-grandfather was born, traveling at 60 miles per hour behind a steam locomotive was the outer limits of technology and Edison had yet to invent his electric lamp. My grandfather was eight years old when the Wrights took off at Kitty Hawk. When my mother was born, general purpose computers in the sense we would use today were 30-odd years into the future. When I was born, human beings had yet to venture into space (historical side note: the Nazis were the first to launch a man-made object into space: the V2 rocket).
Every one of these mileposts was passed many years ago and the pace has constantly accelerated -- almost exponentially, it seems.
Given all this, I just know that although I probably won't live to witness it, the day will come when you will be able to look up toward the ceiling in your home, ask "Computer: where are the children?", and get an intelligible and accurate response.
Add your comments