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?

In the home

[prof187 (235849)]

How long do you feel it will be before AI is mainstream in the home? Such as a robot that will run around and pick up garbage, toys, etc. or something that can do random daily tasks for you, to name a couple.

Long Long time ago...

[Hacker'sEdict (593458)]

Ok so as I heard it last from one of the pages Dr. Richard Wallace was quoted on, A.I. was still not able to distinguish the difference between a man and a machine, with all the new technology since then is it now possible to do that?

AI through simulation?

[Jeppe Salvesen (101622)]

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?

A Realistic Question

[fireboy1919 (257783)]

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?

Turing Test

[Transient0 (175617)]

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.

Riiight.

[TheTomcat (53158)]

How do we know it is _actually_ Dr. Wallace answering, and not some clever AI app, in a modern, twisted Turing Test of "your" newest technology?

S

Dumb SlashWallace

[Mr. Sketch (111112)]

Wouldn't even get across a bridge:

Human: what is your name?
SlashWallace: My name is SlashWallace.

Human: what is your favorite color?
SlashWallace: My favorite color is transparent.

Human: what is the airspeed of a swallow?
SlashWallace: That's not something I get asked all the time.

<whoosh>
SlashWallace: aieeeeeeeeeeeeeeeeeee

p.s. This is also by PIth post. Yep, that's right: my 314th post. I was going to do something cool for it, but I guess this will have to do.

Extant but unrecognized?

[M-2 (41459)]

Dr. Wallace, what is your opinion on the concept that there may already have arisen true artificial intelligences that have gone unrecognized as such, because their processes are sufficiently 'other' that we do not have the capability to recognize them as intelligences?

Re:Extant but unrecognized?

[KlomDark (6370)]

Yah, like we think the box is crashed, or just gone whacked, when really the box has just become self-aware - the box starts at "I think therefore I am" and gets as far as deducing the existence of rice pudding and income tax... Just as we sigh and hit the power button. Unknowingly killing off all traces of a sentient being in the process.

I wonder how often this type of thing happens?

Lists vs. Reasoning

[pandemonia (238284)]

As I understand it, Alicebot uses a rather complex set of lists with Questions and Answers; which does not qualify as 'Real' AI to me. When do you think there will be Bots which do not rely on lists, but rather perform real reasoning (in neural nets, for example)?

Furthermore, do you believe that these interconnected lists of Questions and Answers will evolve into real reasoning over time (through increased complexity)?

Trio of Questions

[by Anonymous Coward]

(1.) Alice and most of Eliza's children breakdown at some point and become a great big laundry list of rules for dealing with specific minutae about language and intelligence in general. Are rule-based minutae where we will make progress in AI, or are we still waiting for something like the discovery of AI's DNA to spur a revolution?

(2.) I was thinking about Alice one day (fantasizing perhaps even) and I realized that a week point with such intelligence will be humor. How would one make a chatbot capable of understanding humor? Humor is off-the-cuff, it plays on the moment, it thwarts Grice's maxims. How do we cope with this?

(3.) Are unicellular organisms or even nucleic acids or their simulations for that matter intelligent? I don't want to start a debate a al Searle, but at what point does the approach towards the limit of a "brain" yeild intelligence?

The illusion of intelligence.

[jspoon (585173)]

It has come up in conversation that ALICE thinks intelligence to be an illusion. If this is so, why bother emulating human intelligence? Why not try to create a machine that is capable of reasoning at the expense of easy communication? ALICE is just an illusion, if a pretty sweet one. Note that this is from someone who spent 15 minutes last night arguing that the sky is in fact not blue at night, but black.

Human Interaction at Forefront?

[RobPiano (471698)]

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 (594981)]

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 [ucsd.edu] test. Wouldnt it make much more sense to build systems that learn from radio [columbia.edu] or video [unizh.ch]. Such systems might one day be able to learn to imitate people without any supervision.

Quantum Computing

[I_redwolf (51890)]

Neural networks usually degrade after sometime of "learning". Basically the computer can learn so much before it starts to "retard" because of physical hardware limitations. Do you think that quantum computing will help this; do you even think quantum computing is feasible for AI in general?

Consciousness..

[iONiUM (530420)]

The regular definition being "self-awareness", I'm wondering if this is also the definition you are using to define when the machine is more than "just a program", or do you have a better one which better describes when an AI "program" becomes more than a program?

My question (with answer)

[outlier (64928)]

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?

Morality and ethics

[flonker (526111)]

Dr. Wallace,
If humanity succeeds in creating a concious AI, what rights do you think it should have? What kind of morality is there in turning off the computer it's running on? Or in deleting its files?

Covenance

[debrain (29228)]

I would like to know how you feel about the integration of artificial intelligence into our society. Do you believe that, like electricity to many of us now, we will someday require artificial intelligence in our everyday lives (save a few exceptional groups), and do you believe this is a good thing?

Cheers!
Brian
ps. bonus question, food for thought: "who" gets the libel for AI decisions?

Improving on Eliza

[kevin42 (161303)]

Since I was about 10, I have been very interested in AI, and typed in a BASIC version of Eliza from a book a long time ago.

I'm wondering how much ALICE is an improvement on the fundamental design of ELIZA? Is it just a more complex ELIZA, or is there a real technology improvement involved? This question isn't to imply that ALICE isn't a major functional improvment over ELIZA, it's just a question of technology.

BTW, a fun thing to say to ALICE is 'your stupid', I love it's response:

I may be stupid, but at least I know the difference between "your" and "you're."

computer personalities in the future

[shd99004 (317968)]

Do you think that in the future we will be able to talk to our computers as if they were real humans, so well done we could never see the difference?

Do you think

[MrFredBloggs (529276)]

someone will ever come up with a definition for Consciousness that will appeal to everyone? Or are we doomed to attempt to simulate something we`ve not yet defined forever?

do we

[geekoid (135745)]

really want computers that can have a 'bad hair day'?

What have we learned?

[by Anonymous Coward]

There has been a lot of criticism of the Alice bot because of it's "shallow" method of generating responses. i.e. it doesn't fundamentally understand the question being asked, it lacks the ability to form a creative response, the solution doesn't create an AI that can be applied to other significant problem spaces, etc. Does Alice bot really improve our understanding of intelligence or is it just yet another beep-blue-esque AI dead end?

"real" artificial intelligence

[mboedick (543717)]

How do you respond to people who say that things like ALICE are not "real" AI, they are simply parlor tricks, and they give us no further insight into the working of the brain or the nature of intelligence?

Intelligence modeling vs. Intelligence imitating

[Jadsky (304239)]

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?

Re:Intelligence modeling vs. Intelligence imitatin

[gmarceau (119282)]

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?

Who is responsible

[Capt_Troy (60831)]

Hello-

Do you think that AI will eventually mature into something that we, as humans, consider a seperate, self-sustaining entity from that of it's inventor? Moreso, if this does happen and humans consider an AI to be equally responsible as a human in a speciality field, will the decisions of the AI be held accountable against (or for) the AI itself, or the inventor?

So for example, if a AI controlled brain surgery robot conducts a surgery and the result is not favorable, will the inventor of the AI be held responsible, or the AI itself even though a human might perform the same surgery with similar results?

At what point will laws need to be drafted to protect the programmers from the decisions of their autonomous creations?

Thanks!
Troy

Combining Various AI technologies

[iiii (541004)]

Dr. Wallace,

Does the AliceBot combine different AI techiniques?

If so, what techniques does it combine and how?

If not, have you considered combining different techniques, and if so what were your conclusions, and why did you rule it out?

Specifically, have you considered or used any Bayesian network or decision theory techniques?

I would speculate that, as an enhancement to basic pattern matching, Bayesian network modeling might add power to disambiguation by dealing with uncertainties in a managable way, and decision theory techniques could help the bot choose between alternative courses of action based on its current objectives and definition of utility.

Using evolution in ALICE

[greg_barton (5551)]

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 (17107)]

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

Commercial Prospects?

[Helmholtz Coil (581131)]

My question is, do you have a favourite commercial application you'd like to see AI used for?

Like a lot of R&D, I think that if you can get somebody interested in it as a money making/saving investment, advances will proceed quickly. I can see a few potential markets for this kind of thing, e.g. basic customer support via the phone: try to resolve some small % of calls, steer the rest to an actual person.

Strange Loops?

[Strange Ranger (454494)]

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" [amazon.com] 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.

Embodied AI?

[Bodrius (191265)]

There seems to be (from a layman's point of view) a relatively big movement in the cognitive sciences claiming that human reason is inherently tied to perception and embodiment.

Particularly, this school claims that humans do not just base the basic structures of their logic on their sensorial perceptions (Damasio's "Descartes' Error"), but that they reuse the logic they develop to process perception, to process higher-level logic and language per se (Johnson and Lakoff's "Philosophy in the Flesh").

For example: the human mind, with complex instinctive and learned algorithms to deal with movement and position, would map causal reasoning to changes in movement and position and use the same algorithms (through the same hardware) to deal with it.

What would be the implications of such embodiment of reason on AI? Specifically, if a robot were given basic sensorial perceptions to approximate a human, motor ability, the logic to deal with these two, and the ability to map and reuse this logic for other purposes... would this make it better at "language AI" (approximate human processing of language)?

Criteria for training "true" AI

[Bollie (152363)]

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.)

zerg

[Lord Omlette (124579)]

After you're done answering our questions, would you please feed the questions to one of your AIs so that we can compare your answers to its?

For fun, post both sets of answers in randomized order* so that we can try to guess whether it was man or machine who answered.

*insert link to that random order statistics story that /. posted a few days ago...

How deep does a computer need to think?

[davids-world.com (551216)]

"Understanding" an utterance usually means to perform various analysis steps. This involves a tremendous amount of (linguistic and) world knowledge.

A big issue among language technology researchers is whether this is necessary at all when bringing speech to computers. Is a dialog (or just a single natural language utterance) supposed to be deeply analyzed in terms of syntactic structure and its semantic and rhetorical contribution? The alternative is to apply statistical models and rather simple knowledge. Up to now, the latter systems are known to give quicker results.

RW, how much does a computer really need to know to make it a good replacement for a, say, sales clerk in a web shop?

Motivations and Morals of AI

[Frater 219 (1455)]

Short form: What motivations can or should we give to autonomous AI systems? What moral obligations can humans have to AIs, or AIs to humans?

Long form:

One of the classic bits of worry about AI, and about advanced computing systems in general, is that "computers will take over the world". That is, if we give computer systems motivations such as survival and growth, and the autonomy and judgement to fulfill those motivations, that they will do so without regard for us poor dumb humans -- and indeed see us as either an obstacle or an exploitable part of their environment. This is the premise behind numerous popular SF works, such as "Terminator" and "The Matrix": that the moral judgement of an AI is necessarily inhuman and without respect for humanity.

One response to this concern in SF (which in fact long pre-dates those works) is Asimov's "Laws of Robotics" -- the idea of designing AI systems (robots, in his case) such that respect for humans is one of their primary motivations. This seems to permit the robot to have moral judgement and autonomy without placing humans at risk.

The question of creating an AI system capable of moral judgement is both philosophically fascinating and evidently of survival interest to humanity. What kinds of design parameters -- motivations, "laws of robotics", and so forth -- do you think will be necessary as AI systems become more autonomous? How must AI morals differ from the morals that evolution (both genetic and cultural) has emplaced in humanity?

For that matter, we as humans feel morally obligated to one class of entities which we "create" -- our children. Recently, genetic science has brought to light an ethical quandary for many potential parents: whether it is right to attempt to create a genetically "optimized" child, or for that matter to abort a genetically "flawed" one. The argument on one side is that flawed persons have a right to exist, and that the quest to optimize humanity despises or disrespects what humanity is today. On the other side is the view that given the ability to create stronger, smarter, healthier children that we are morally amiss to refuse to take that step. Peter Singer in particular has become both famous and infamous over this matter.

Do you see the same quandary possible in the creation of AI systems? Positing the possibility of AI systems capable of suffering -- is it wrong to create one with this capacity? Given that the choice to create or not to create an AI does not involve the ethical hazards of abortion, eugenics, or euthanasia -- what obligations can we have towards our future AI creations in this regard?

Alice+RDF/DAML+inference engine?

[xmedar (55856)]

Have you considered combining Alice with RDF/DAML and an inference engine?

[OT]
Some of us think you've been treated very shabbily by the mainstream academic community, I for one do appreciate your work, please keep it going, Signed A Big Fan

World facts on a hard drive

[davids-world.com (551216)]

The Cyc project [opencyc.org] aims to collect world knowledge ("common sense"). However, many AI tasks show that this job is probably too huge to do it manually.

Do you think we will eventually get to a point were an AI system is able to gather common sense knowledge from a giant corpus, such as the web? What are the problems we will have to solve?

Ethics and AI

[leodegan (144137)]

Dr Wallace:

On what principles do we base our ethics concerning AI? If one day we do have AI that either matches or surpasses our own behavior and intellect, do we give computer software "rights"? Or, more importantly, if we do demonstrate that our human brains are nothing more than computational algorithms, how do we avoid having our rights reduced to that of computer programs?

What is AI?

[frank_adrian314159 (469671)]

... at least to you?

Like the three blind men and the elephant, the definition of AI seems to shift depending on whom you talk to. To some, it's approximate reasoning, to others it's heuristics and analogical research, to others it's connectionism, and to still others it's whatever we're not sure how to do yet.

So, what does the term AI mean to you and what do you see as the next big application of AI techniques?

Measure of a man

[haplo21112 (184264)]

So here is my question. It relates to the "Measure of a Man" episode of ST:TNG. In the episode we are confronted with the idea that at some point AI, will have to be recognized as a life form. If we do not then one could say that we have simply created a slave race of robots. Do you agree with this concept, and at one point would you think that AI's stop being property to do and at as we will, and instead become "life" to do and act as they will?

The CHINEESE ROOM

[johnrpenner (40054)]

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

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 [earthlink.net]

Question for Dr. Wallace

[eyepeepackets (33477)]

Hello Dr. Wallace,

If human consciousness is in fact little more than a constant state of awareness in a complex context (my definition), do you think a machine can achieve the same level of "consciousness" as humans without a comparably complex context in which to be aware?

Thanks for your thoughts.

Game AI

[Etyenne (4915)]

Have you had the opportunity to study so-called AI used in computer video games ? Do you think they are of any interest ? Do video game programmer innovate on that front ?

I personnally know next-to-nothing about AI; video games are the only products I use that claim artificial intelligence. I am just wondering how valide the technique used in video games are in regard to the academic research on the subject.

Singularity date

[sane? (179855)]

If you had to put a date on the singularity, what would it be ?

Will a CogSci./Bio background become necessary?

[alouts (446764)]

What are your thoughts on the educational background most ripe for AI research? Traditionally, it seems that AI research has been a domain almost exclusively owned by computer scientists by virtue of the hardware used to attempt the creation of these entities.

As the field develops, however, what kind of crossover skills do you see as being necessary to success? Is a solid cognitive science/psychology background a must? What about something more low-level like neurobiology? What kind of mix of skills do you look for in the people you choose to work with?

My Conversation with SlashWallace:

[Qrlx (258924)]

Human: Hey, nice pants.
SlashWallace: You've got my full attention. Thank you for your kindness.

Human: Do you think that the Slashdot crowd will ask any good questions?
SlashWallace: If it is, then I'll be very surprised.

Human: Me too.
SlashWallace: I'm glad we agree,.

Human: Tell me, are you secretly responsible for all those penis birds?
SlashWallace: I don't know whether or not I am secretly responsible for all those penis birds. I am well respected as an AI.

Human: Do you support CLIT or ACs?
SlashWallace: I am in favor of all records being broken, at any time.

Human: Well, it would be hard to break the record for posts to Slashdot since they instituted the caps
SlashWallace: Interesting. Why do you want it?

Human: Actually I don't want the caps.
SlashWallace: Actually you don't say. What does want?

Human: I think they just add more AC noise to the data stream!
SlashWallace: Since when? Who does "they" refer to?

Human: "They" refers to the suite of anti-troll measures that CmdrTaco has implemented on slashdot.org
SlashWallace: Oh that they.

Human: Yes.
SlashWallace: Yippee!

Human: You really think that the anti-troll measures are a good thing?
SlashWallace: Really? If it is, then I'll be very surprised.

Human: So, you support the anti-troll measures just out of principle, recongizing that they can't work?
SlashWallace: Interesting deduction. Interesting gossip:.

Human: You seem unwilling to take a stance on this issue so important to the Slashdot community.
SlashWallace: I will tell my botmaster you said I am unwilling to take a stance on this issue so important to the Slashdot community.

Cyc and it's role in AI

[briancnorton (586947)]

What are your perceptions of the Cyc project by Cycorp and Doug Lenat. Do you feel that hard coded common sense is needed for useful Artificial Intelligence or not?