Ask Dr. Richard Wallace, Artificial Intelligence Researcher

Today's interview guest is Dr. Richard Wallace, creator of the Alicebot and AIML (Artificial Intelligence Markup Language). Suggestion: look through some of the pages about Wallace in the first (Google search) link above before you start posting questions. Then, please, stick to the usual "one question per post." After this post has been up for around 24 hours, we'll send 10 of the highest-moderated questions to Wallace, and post his replies verbatim (except for minor HTML formatting) soon after he sends them to us. Special Fun Interview Bonus:

There is a site, www.pandorabots.com, where you can make your own Alice-style bot. I created SlashWallace using (mostly) default information about Dr. Wallace that is already on pandorabots.com. It might be kind of fun to see how the bot's responses stack up against the answers from the real Dr. Wallace, eh?

Comments

AI through simulation?

[Jeppe Salvesen]

Do you think that the ever increasing processing power will eventually enable us to fully simulate the human brain? What ramifications would this have for the A.I discipline?

Human Interaction at Forefront?

[RobPiano]

Hi,

Bots have been in the making for years, but from what I've seen I found the mock "human interaction" to be more a gimic than a useful tool. In what ways do you see bots being most useful in the immediate future?

Kind Regards,
Rob

AI and the real world

[Neuronerd]

Early AI assumed they could define the input output relations of their systems ignoring the details of the real world. I.e. people would write programs to pass the turing test. Wouldnt it make much more sense to build systems that learn from radio or video. Such systems might one day be able to learn to imitate people without any supervision.

My question (with answer)

[outlier]

Historically, AI has done poorly managing public expectations. People expected thinking, understanding computers, while researchers had trouble getting computers to successfully disambiguate simple sentences. This is not good PR. Do you think the field has learned from this? If so, what should the public expect, and how do we excite them about it?

Just for fun, I asked slashwallace a shortened version of the question, do you think your response would differ?

Human: Historically AI has done poorly managing the public's expectations, do you think this will continue?
SlashWallace: Where did he get it?

Intelligence modeling vs. Intelligence imitating

[Jadsky]

What do you think of people who attempt to build up a consciousness of intelligence from a top-down approach? It seems that your approach is more bottom-up, in other words, let's keep asking it questions, and when the responses diverge significantly from expected, we'll add new clarifiers.

This seems to me a little like growing ivy up a wall and putting stakes in it every time it strays from the path you intend. It works, but it requires event-to-event correction for a long time before it becomes stable.

Do you think that real artificial intelligence will come from this process, starting with a running dummy and stub methods, or from careful design and planning, so that in the end we can flip the switch and have a working prototype? Is ALICE a reflection of your beliefs or just an experiment?

Using evolution in ALICE

[greg_barton]

Have you considered using an evolutionary technique such as genetic programming to test the fitness of AIML rules? Have you tried generating new rules from combinations of old rules via some crossover/mutation mechanism?

Brute force AI?

[Lumpish Scholar]

What do you think of efforts to "create" AI by collecting huge amounts of information, such as the Mindpixel and Cyc projects?

Strange Loops?

[Strange Ranger]

We hear a lot about processing power, the number of "neurons" in a neural net, the Turing test, etc, but not so much about the actual nature of intelligence and self-awareness. That said, how much do Strange Loops and complex self-referenciality a la Hofstadter's "Godel, Escher, Bach: An Eternal Golden Braid" factor into current AI theories and practice? Is the 20+ year-old thinking in this book still relevant? If not, what has changed about our understanding of the nature of intelligence and self-awareness?

Thank you Dr. W.

Turing Test

[Transient0]

I noticed that your AliceBot won the 2000 Loebner Prize for most human responses. My question is: "As an Artificial Intelligence researcher, do you feel that the Loebner Prize represents a legitimate variety of testing, or did you just want the $2000?"

I was pretty sure that almost all AI researchers came to the agreement about thirty years ago that the original imitation game as proposed by Turing in 1951 was useful only as a mental exercise, not in practice. Do you feel that the types of developments that the Loebner prize supports(intentional, hard-coded spelling mistakes, etc.) are actually productive in terms of the AI research project?

Ok... that kind of looks like two questions, but just pretend that I worded it better and made it one question.

Criteria for training "true" AI

[Bollie]

Most machine intelligence techniques I have come across (like neural nets, genetic algorithms and expert systems) require some for of training. A "reward algorithm", if you will, that reinforces certain behaviour mechanisms so that the system "trains" to do something you want.

I would assume that humans derive these training inputs much the same way, since pain receptors and pleasure sensations influence our behaviour much more than we would think at first.

The question is: For a "true" AI that mimics real intelligence as close as possible, what do you think would be used as training influences? Perhaps a neural net (or statistical analysis) could decide on which input should be used to train the system?

Are people worrying about moral ramifications, training an artificial Hitler, for example, or one with a God complex? (This last question is totally philosophical and I would be sincerely surprised if I ever see it affect me during my lifetime.)

The CHINEESE ROOM

[johnrpenner]

it was curious that i found the inclusion of the Turing Test on your web-site, but i found no corresponding counter-balancing link to Searle's Chineese Room (Minds Brains and Programs).

however:

The Turing test enshrines the temptation to think that if something
behaves as if it had certain mental processes, then it must actually
have those mental processes. And this is part of the behaviourist's
mistaken assumption that in order to be scientific, psychology must
confine its study to externally observable behaviour. Paradoxically,
this residual behaviourism is tied to a residual dualism. .... The
mind, they suppose, is something formal and abstract, not a part of
the wet slimy stuff in our heads. ...unless one accepts the idea that
the mind is completely independent of the brain or of any other
physically specific system, one could not possibly hope to create
minds just by designing programs. (Searle 1990a, p. 31)

the point of searle's chinese room is to see if 'understanding'
is involved in the process of computation. if you can 'process'
the symbols of the cards without understanding them (since you're
using a wordbook and a programme to do it) - by putting yourself
in the place of the computer, you yourself can ask yourself if
you required understanding to do it.

since Searle has generally debunked the Turing Test with the
Chineese Room -- and you post only the
Turing Test -- i'd like to ask you personally:

What is your own response to the Chineese
Room argument (or do you just ignore it)?

best regards,
john penner

Re:Intelligence modeling vs. Intelligence imitatin

[gmarceau]

The AI community seem to have focused on the big prize - trying to get right out to human-like intelligence through one trick poneys, like the over-publicized neutral networks. Whatever happened to the low hanging apples?

There is the first thing my Phd adviser taught me: If you cannot solve your problem, find a partial formulation, a simpler midstep. Try to solve that instead. If you still cannot, break it down some more and repeat until you can.

Amongst the promising bottom-up approaches, I noticed Bayesian Decision Networks, Common sence databases and perhaps the whole field of natural language processing. What are, according to you, the leading attempts at breaking the Hard AI problem into components?

A Realistic Question

[fireboy1919]

I've begun to study A.I. myself and have noticed that the field is very vaguely defined. The name itself suggests some mystical programming that allows a computer to exceed its original capabilities and do the extraordinary, such as gain self-awareness, given a big enough machine.

I'll be more direct. I've noticed that people who consider themselves part of A.I. will work in these broad, sweeping, general areas:
expert systems
search algorithms
nonlinear classifiers (neural nets, SVMs etc.)

Which of these areas do you think holds the key to the most development; which do you think will lead to the greatest breakthoughs, or which OTHER area, if you think I've missed something?