Biomedical Engineering Blog

Biomedical Engineering

The Biomedical Engineering blog is the place for conversation and discussion about topics related to engineering principles of the medical field. Here, you'll find everything from discussions about emerging medical technologies to advances in medical research. The blog's owner, Chelsey H, is a graduate of Rensselaer Polytechnic Institute (RPI) with a degree in Biomedical Engineering.

Previous in Blog: Predicting Your Weakness   Next in Blog: Engineering Synthetic Minds (Part 2)
Close
Close
Close
6 comments

Engineering Zombies (Part 1)

Posted March 24, 2008 12:01 AM by Steve Melito

"We're trying to engineer zombies. That's what we're trying to do," explains Dr. Selmer Bringsjord, director of the Rensselaer Artificial Intelligence and Reasoning Laboratory (RAIR). Bringsjord, who also chairs the Department of Cognitive Science at Rensselaer Polytechnic Institute (RPI), is the author of tracts such as "What Robots Can and Can't Be" and "In Defense of Impenetrable Zombies". Last week, Dr. Bringsjord met with CR4's frankd20 and Moose at the Troy, NY offices of GlobalSpec, the company which powers CR4. For over an hour, Professor Bringsjord answered questions about his current research, and discussed the promise and perils of artificial intelligence (AI).

Technical Philosophy

Selmer Bringsjord is an imposing figure, a man who towers above his peers in terms of physical size and intellectual achievement. Yet he is also extremely approachable, a college professor who answers email from strangers like us, and regales his students with tales of lost luggage. A graduate of the University of Pennsylvania (B.A., magna cum laude, 1981), Bringsjord earned a Ph.D. in Philosophy from Brown University in 1987. There, he read Eugene Charniak's "Introduction to Artificial Intelligence" and took a course by the same name. Eschewing the work of the great philosophers, Bringsjord devoted himself to what he calls "technical philosophy in all its forms." The link between logic and computer science is a strong one, and the philosopher's dedication to the former had implications for the latter.

IBM, RPI, and CCNI

Today, Selmer Bringsjord leads a project called "Engineering Cognitively Robust Synthetic Characters." Formerly known as RASCALS, an acronym for "Rensselaer Advanced Synthetic Architecture for Living Systems", Bringsjord's research puts the RPI professor in front of one of the world's fastest supercomputers. As reported last month in CR4, Rensselaer Polytechnic Institute has teamed up with IBM and New York State to form a $100-million partnership, creating the world's most powerful university-based supercomputing center. Located in the Rensselaer Technology Park (which is also home to GlobalSpec and CR4), the Computational Center for Nanotechnology Innovations (CCNI) boasts a series of IBM BlueGene/L systems which contain a total of 32,768 Power OC 440 700 MHz processors.

The Turing Test

So why does Selmer Bringsjord need so much computing power to conduct AI research? It's a simple story really – at least for a human to tell. During the 1950s, an English logician named Alan Mathison Turing devised the Turing Test, an exercise in which a person engages a machine in natural-language conversation. In order for the machine to "pass" the test, the person must mistakenly conclude that the machine (a form of artificial intelligence) is another human being. Turning a machine into a storyteller isn't so simple, however. "Any AI at all is a form of a zombie", Dr. Bringsjord explains. AI can be "sophisticated", but artificial intelligence is devoid of emotion. Consequently, "the characterization is very shallow" in machine-told tales. AI researchers grow frustrated, and human test subjects remain unconvinced.

That's where supercomputing comes in.

Editor's Note: Click here for Part 2 of this four-part series. Part 3 and Part 4 are also available.

Steve Melito - The Y Files

Add to Technorati Favorites

Reply

Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.
Guru
Engineering Fields - Marine Engineering - New Member

Join Date: May 2007
Location: Australia.
Posts: 1642
Good Answers: 81
#1

Re: Engineering Zombies (Part 1)

03/25/2008 4:45 AM

This is a difficult subject to engage in, not being fully familiar with it. But I think it is a subject that many would have though over in the past, if I was to design a robot how would I go about it approach.

As the A1 robot design gives no details as to how it engages in story telling other than the size of the computer, and as I assume the article is about that particular standard?

I would like to make just a few remarks along the lines, how would I go about it? There is the problem of recognition and evaluation, one assumes Boolean logic is the computer approach, something I do not fully agree with, but rather emotive self concern, how do you give a computer emotions? How do you priorities recognition? Recognition is a complex subject, something I do not think of it as a one dimensional subject, nor do I think it is Boolean, but one of priority?

Just a short note to get the ball rolling.

Regards JD.

Reply
Guru
United States - Member - New Member Technical Fields - Technical Writing - New Member Popular Science - Weaponology - Organizer Hobbies - Target Shooting - New Member Engineering Fields - Nuclear Engineering - New Member

Join Date: Mar 2005
Posts: 3464
Good Answers: 32
#2
In reply to #1

Re: Engineering Zombies (Part 1)

03/25/2008 10:00 AM

Good to hear from you, JD. Thanks for getting the conversation started. AI researchers cannot "program" emotions. That's part of the reason why synthetic characters are zombies - they can't feel anything. Hypothetically, you could have a conversation in which a human says "I'm happy today" and the AI says "I'm happy, too". But the AI doesn't feel happiness. It's just been programmed to respond in that way when a "I'm happy today" statement is made.

A subsequent part of this interview will discuss some of the implications for "emotionless" AI, especially with regard to weaponizing synthetic characters.

Reply
Guru
Engineering Fields - Marine Engineering - New Member

Join Date: May 2007
Location: Australia.
Posts: 1642
Good Answers: 81
#3

Re: Engineering Zombies (Part 1)

03/25/2008 6:26 PM

In MHO I disagree that emotions cannot be programmed. I think at least the basic building blocks and structures can be formulated. What is the history regarding efforts to accomplish this, or has it been dismissed outright?

Regards JD.

Reply
Guru
United States - Member - New Member Technical Fields - Technical Writing - New Member Popular Science - Weaponology - Organizer Hobbies - Target Shooting - New Member Engineering Fields - Nuclear Engineering - New Member

Join Date: Mar 2005
Posts: 3464
Good Answers: 32
#4
In reply to #3

Re: Engineering Zombies (Part 1)

04/01/2008 11:25 AM

Sounds like we could have a good debate about this, JD! Certainly, emotions can be "formulated" in that sense that drugs can alter the levels of neurochemicals such as serotonin. But is the administering of medications really "programming"? I'd say not.

I don't know the history of trying to program AI with emotions, but am hoping that others who may know more will join us in this discussion.

Reply
Guru
Engineering Fields - Marine Engineering - New Member

Join Date: May 2007
Location: Australia.
Posts: 1642
Good Answers: 81
#5

Re: Engineering Zombies (Part 1)

04/01/2008 6:30 PM

Others with more knowledge will need to expand, but when I refer to building blocks and formulation, what I have in mind is the following.

First there is the sensory input, that requires recognition, this has two associated blocks, what I call the auxiliary, and grading, the auxiliary is the associated muscular movement, and grading is simply sensory like and dislike.

Secondly formulation, this I think of as the primitive emotion of survival, collective behaviour to provide the necessities of survival, food. The thought of success evoking elation and laughter, failure evoking fear and screams, these two actions conveying a primitive message to a group.

So putting this together how does it work? First a sensory input is graded and formulated, then depending on the result stored accordingly, this is similar to C++, objects. They are contained in there separate groups, so a input associated with that group only evokes responses from that group. And these groups are tied together though the auxiliary.

Thirdly, auxiliary formulation. This is the non emotive part of the brain that structures responses? This is a friendly block, that can access all other blocks, restructure and test concepts, the stored results are the thoughts that are communicated to others, the result passing to a formulated group? This area has the responsibility to respond or not respond. Something possibly related to experience and its own history. Pain and pleasure?

Just a few thoughts, Regards JD.

Reply
Guru

Join Date: Oct 2010
Posts: 1294
Good Answers: 35
#6

Re: Engineering Zombies (Part 1)

06/09/2011 1:02 PM

So, October, 2008 has come and gone--any feedback on what happened?

Reply
Reply to Blog Entry 6 comments
Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.
Copy to Clipboard

Users who posted comments:

jdretired (3); rhkramer (1); Steve Melito (2)

Previous in Blog: Predicting Your Weakness   Next in Blog: Engineering Synthetic Minds (Part 2)

Advertisement