AI Zone Admin Forum Add your forum

NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

Intuition and Empathy
 
 
  [ # 31 ]

There are painting robots who can draw from life, adopt other people’s styles and so create original works, but nobody is even suggesting they are intelligent. Prejudice is the obstacle. In my opinion creativity does not equal intelligence, though it may be an expression of intelligence. a number of alternative tests have been suggested. I think most of them will be met with the same denial as the first chess computer.
If an AI of my creation would have to prove itself intelligent, I would simply have it reason with the people about whether it was intelligent, and whether such a thing was provable. If it truly were as intelligent as me than it should probably end up convincing the judges that if it can not be proven for humans, then it can not be proven for AI either.

 

 
  [ # 32 ]
Steve Worswick - Jan 12, 2016:
Amanda June Hagarty - Jan 12, 2016:

Steve, I think that perhaps some people don’t stick with AIML long enough or really explore it’s full potential to see the complexity it is capable of creating.

This is exactly the issue. People play with it for a while and then give up. Even Andres says he ONCE tried it. If I were to try playing piano and made a noise on my first attempt, this is normal. I certainly wouldn’t start claiming what it can or cannot do.

Andres - If your bot had 2,000 categories and was as bad as you say, you need to look at what you were doing than writing the whole language off. I have around 300,000 categories and my bot usually replies in less than a second.

Hi Steve, you have all my full respect, and may be a hell of a AIML programmers, or whatever. But let me introduce some concepts here which might be misunderstood by many readers/writers here, no regrets.

As I can barely understand AIML, a category is just a path in the graphmaster, which connects among situations, or partial matches which gets simplified over and over until a full “parsing” in terms of the graphmaster is achieved.

But a category set cannot be compared by far to a single true-parser, even the tiniest chunker, whichever kind it is. Almost all Natural Languages have non-regular grammars, so they must be handled by a mixup of statistics and deterministics, like a pull-down automata, governed by relations, as well as a disambiguator, pos tagger and all the crap you surely know, or can find out in wikipedia as well. I don’t want to go too technical, but not a LL nor any LR parser won’t work fine on human grammar, even GLR parser which complexity if more than O^3 can’t cope fully with irregular human grammar. (I tried it also) only some dependency grammar and chunking works fine with unstructured dialog-like conversations.

for example a AIML section-pattern like
[subject] [direct object] [predicate] is barely impossible, and in a parser (or my DDL language) you can do this in a breeze, on single line!

As showed above, any single section grammar oriented parser can be equivalent to a unlimited number of categories, and the power of a graphmaster-like path below this is almost unlimited. As you add categories to a AIML engine, the more patterns you have at one single point, and worse if they are nested; the slower render the engine, because the system can go into a huge testing loop. Many engines fought this, adding catching (I must confess, on certain loops I also do this) but it is not enough to assure a decent speed. (respond under a few milliseconds, on a server).
Also the complexity to handle all this categories who embed the responses is too big, and exceeds normal flow of response of an agent.

You cannot set up a selector to the next response, in any insulated section of the conversation, based on (let’s see.. a 300 documents text-file set), and drag the conversation on the subject of any of them using a statistically learned classifier like a SVM-based on semantic-vector-space similarity, which has a better than 95% quality (F-score ~ precision and recall) and based upon the success of the conversation you can make this section to learn better any new pattern it might catch!

Also all the AIML responses are somehow “canned” or built out of relations found on the path to the patterns, but you can hardly do something like “read 100k documents, and decide you have been talking with a user about “similar stuff” to the content of a tiny sub-group of them, and find out what section of which document you were talking about, without writing tons of AIML. And then do the “magic” like answering some user’s questions, deducted by a true reasoning engine, which is embedded in the specific sub-issue/problem you were talking about. To build a AIML that can respond questions about a continuously changing (even learning) database is (at least for me) impossible, unless you re-load continuously automatically generated AIML from the data, which is unpractical.

I bet you to solve a simple exercise, ¿can any write a AIML section that understands full math?, lets see all the 4 operations (+ - * / ) of any order and length, even with parenthesis and nesting, regarding the priority of the operations? - As I can see it this may be too complex (I tried unsuccessfully, it and saw/thought this might be impossible by the model itself)
And provided you do it, ¿how many times I need to copy this sections to use it over and over, in different sections? like a subroutine.

¿Can a AIML system with 300k categories, talk simultaneously to 1000 users without collapsing?
As I see each ‘user-talk’ needs to be “booted” (load the categories) filling the server memory up. Am I right or wrong?

Despite of that, the kind of things you can accomplish with AIML is astonishing, but as well as you scale in complexity the number of AIML lines you need to create does not scale linearly with the problem, it scales potentially (dont know the index) but it’s not 1.

I also apologize to you all, about my English writing, which might not be as expressive and rich as a native English speaker, but I try as much as I can. I learned English at age 6, got my other 2 mother languages from the crane, and my last 3 languages at age 21.

I am not a detractor, but the single approach of parsing as well as responding, and decision making of AIML makes it, as I can see, a hobby language, nothing more, nothing less.

 

 

 
  [ # 33 ]

Don, well being able to prove one’s own intelligence is a little bit like being able to prove that one is not just a brain in a jar receiving nerve impulses that simulate existence. Though I see what you are saying. If the AI doesn’t need to “prove” but rather just needs to come up with reasoned arguments, then in a way the ability to come up with a reasoned argument is in itself a proof, and a great way to measure intelligence! Thanks for playing along!

In terms of creativity, I was thinking not of expressive creativity like painting, but perhaps inventive creativity (I have been reading a bit since my posts this morning to refresh myself on the subject). And I might give the AI a problem to solve such as:

You arrive at the door step of a friends house and they are not home. You need to leave them a message but you don’t have a cell phone or pen and paper. There in nothing here except a welcome mat on the pavement, a wooden door, a small garden with plants, dirt and rocks. How do you leave a message with only these objects and without coming back later.”

The machine would then have to create its own solution. And if you gave the machine several of these types of situations s you could arguably come up with a measure of its creativity to compare to average people and be able to determine e whether or not the machine had the same creativity as humans.

And no creativity is not intelligence, so this would not be a measure necessarily of intelligence strictly speaking. But as you say the two seem related and proving that a machine can think as well as a human is about more than just intelligence. At least that is what I was thinking.

 

 
  [ # 34 ]

Interesting. I am only starting to work on the AI skill of problem-solving, but that does sound like a good test of intelligence. Much better than something as subjective as art (a more popularly suggested test of creativity). Ravens and dolphins are also considered intelligent for their ability to use tools, at least to some degree.

 

 
  [ # 35 ]

I don’t intend to hijack Amanda’s thread any longer with this but will say the following:

Andres Hohendahl - Jan 12, 2016:

Also all the AIML responses are somehow “canned” or built out of relations found on the path to the patterns

Nope. You can create dynamic responses and answer questions you haven’t hard coded.

Andres Hohendahl - Jan 12, 2016:

I bet you to solve a simple exercise, ¿can any write a AIML section that understands full math?, lets see all the 4 operations (+ - * / ) of any order and length, even with parenthesis and nesting, regarding the priority of the operations?

Already done. I coded some AIML to work out maths up to 9999. As it’s primarily a conversational bot, that was enough for me but it’s perfectly possible to create a loop to handle numbers of any length.

Andres Hohendahl - Jan 12, 2016:

And provided you do it, ¿how many times I need to copy this sections to use it over and over, in different sections? like a subroutine.

You don’t need to copy it at all. Once you’ve written a function in AIML, you just call it with whatever parameters you need.

Andres Hohendahl - Jan 12, 2016:

¿Can a AIML system with 300k categories, talk simultaneously to 1000 users without collapsing?

Easily. Mitsuku was promoted on Kik messenger for a month or so and spoke with around 250,000 users each day. The server barely noticed.

Andres Hohendahl - Jan 12, 2016:

As I can barely understand AIML,

I think this is the crux of your issues. I am happy to help you with any aspect of AIML if you wish.

Amanda - Proving intelligence is extremely difficult. At one time, chess was seen as the ultimate challenge of intelligence, until Deep Blue came along. People then said, oh it can’t be intelligent. The same would happen with your “leave a message” problem. If a computer solved it, people would claim it was some kind of random chance or whatever.

Animals are certainly intelligent and I would say more intelligent than humans. In winter, I used to go to work in the snow and rain while leaving my cat at home staying warm. I’m out working to earn money to buy cat food and toys while my cat lazes around all day. Who is the more intelligent one? wink

 

 
  [ # 36 ]

Don, us it is a common misconception that painting and art are the only kinds of creativity. Many people have made definitions of creativity that include more. I.R. Taylor (1959) proposed 5 levels of creativity: Expressive, Productive, Inventive, Innovative, and Emergenative.

Steve, expecting a computer to do an arbitrary thing such as beat a chess player or fool people into thinking they are chatting with a human will always be quickly discredited afterwards. Such benchmarks are good for pop science mostly and to drive innovation by giving developers something to beat. But proper studies are needed to get closer to proof. I say “closer to” because science never really proves anything. Science is really about what is currently not disprovable. The study which I suggested (really just off the top of my head so its not really all that sound lol) would basically show that either a robot could come up inventive solutions to such problems *at the same rate as the average human* or it could not. Thus offering “proof” that it had the same creative intelligence as a human in that particular level of creative intelligence using Taylor’s definition specifically. And then it’s up to other scientists to make a bunch of attempts to disprove it or poke holes in the method. Public opinion can be whatever it is. If other scientists can’t disprove it or find a flaw in the method it doesn’t really matter what the media says or what people say sitting around their kitchen table wink

And I specifically said *at the same rate as a human for 2 reasons. 1) Not every human will be able to come up with a solution as levels of creative intelligence vary (my solution to that problem, which was a real life event for me that happened when I was 12, was to shove the welcome mat aside and pour handfuls of dirt out into the letters: “mom, please bring my gym shoes to school” lol). 2) You would need the comparison to humans to combat the types of discrediting you are talking about. If the robot simply came up with a solution to problems without being compared to human responses then you would have no way to say that this showed the robot was equal in performance to humans :D

And if you want to talk about the intelligence of other life forms in terms of enslavement of the human race, think for a moment about plant intelligence. Apple trees, wheat, corn, look at how they have enticed us into caring for them, propagating them, improving them. Now that’s a feat of intelligence right there lol.

 

 
  [ # 37 ]
Steve Worswick - Jan 13, 2016:

I don’t intend to hijack Amanda’s thread any longer with this but will say the following:

Andres Hohendahl - Jan 12, 2016:

Also all the AIML responses are somehow “canned” or built out of relations found on the path to the patterns

Nope. You can create dynamic responses and answer questions you haven’t hard coded.

Andres Hohendahl - Jan 12, 2016:

I bet you to solve a simple exercise, ¿can any write a AIML section that understands full math?, lets see all the 4 operations (+ - * / ) of any order and length, even with parenthesis and nesting, regarding the priority of the operations?

Already done. I coded some AIML to work out maths up to 9999. As it’s primarily a conversational bot, that was enough for me but it’s perfectly possible to create a loop to handle numbers of any length.

Steve, can you explain how you need to code this, and be reusable and somehow optimal…

Under my DDL language, you never have to code any parser, it is built in! you simply wait for a number, and whatever the number might be formed of, words, calculus or a clever combination, you get the result at the evaluation part of the pattern.

I really cannot figure out how this could be accomplished with AIML without tweaking the engine itself, and handle the operations almost by hand, setting exceptions to handle priority, etc. A total nightmare, and you are right in something, AIML may code almost anything, but in the same way I think any drawing, as complex as you wish, can be accomplished with turtlegraphics, given enough time and effort. What matters is the speed to code it and hence the time+effort it needs. Jurassic Park could have never been filmed nor sold without powerful computer graphics, but we would still be waiting 100 years or more for the trailer, if made with turtlegraphics.
No offense, please ( I’m joking). smile

I would be enchanted to find how to create a recursive reentrant parser with AIML or a LR parser just with a ambiguous grammar like the one with sum and multiplication.
I am interested in learning from the one who knows more, not destroying the reputation of AIML
So, I am all ears!

 

 

 
  [ # 38 ]

As this is way off topic, I have demonstrated an example of maths and processing BODMAS on this thread:
https://www.chatbots.org/ai_zone/viewthread/2354/

Alaric has demonstrated a simple POS tagger here:
https://www.chatbots.org/ai_zone/viewthread/2353/

 

 
  [ # 39 ]
Steve Worswick - Jan 14, 2016:

As this is way off topic, I have demonstrated an example of maths and processing BODMAS on this thread:
https://www.chatbots.org/ai_zone/viewthread/2354/

Alaric has demonstrated a simple POS tagger here:
https://www.chatbots.org/ai_zone/viewthread/2353/

Thank you!

This way of programming confirms me my theory of “turtlegraphics” vs. AIML, In my humble opinion, there could hardly exist a more complicated, or non-intuitive way to program a conversation or function, using this methodology, you must think on the loops, the re-entrant part, you only have very little resources to tailor the process flow.

I must congratulate anyone like you in being able to think this way, by following this circular and multiple looping paths, as well as remembering them after a while; because I never could find out how to document reentrant and looping code, nor thinking this way.

I think there are other options out there, and I went to create it, mine might not be as fancy and sophisticated as AIML, but it fulfills, let’s say, several hundred of AIML code, with just one or two intuitive-text code. Just that!

If you want to compare two computer languages A and B, to fulfill some task, you cannot say “with A, you cannot build the algorithm, but with B you can”. the comparison is deeper and subtler.
You can say: “A accomplishes this in 3 lines of clearly understandable and intuitive code”
and “B can do this with 30 lines of complicated and reentrant, non-intuitive code”

Ok you can say “intuitive” is perfect to a guy who spent 10 years programming under A, and anti-intuitive might be the other guy who never learned A, uses B and sees A this as crap. It’s all a matter of opinions. but there are several programming principles / patterns: “shorter is better” and “clearer is better” which combined yields “shorter and clearer” is better² and this combination is by far, less error-prone!

The second part I criticize on AIML is the “theory of the graphmaster” whenever you search for a match, you will flow in a NP (Number Permutation) way to find it, under AIML it’s really hard to code and control a noise-words ignoring algorithm, or a smart chunker, who evaluated many combinations and determines the best probability, even you state there is no disambiguation possible for POS tagging, and this is crucial even for English.

My actual and upgraded conclusion is that AIML is a good language built for simplicity, severely limited to build big things and to include new algorithms as “de facto” as well as its methodology is rather archaic but unique, on my humble point of view. It’s like a toy for me, you can build a house to live in just using Lego, but it will be very expensive, time-consuming and complicated as well as many functionalities will be just forbidden. No regrets wink

 

 

 
  [ # 40 ]

Isn’t it about time you made your own thread on this topic Andres?

 

 
  [ # 41 ]

Don’t know how to do this (split the thread), 
But I will not continue to post here, because maybe it’s off-topic as you say. sorry :(

 

 
  [ # 42 ]

No problem. What I suggest you do is if you have any further thoughts on the topic you create a new tread under a coding category and then make your post there. Then copy the URL of that thread and paste it into a comment here so anyone who wants to discuss it further with you who was following over here can easily find your post smile

 

 
  [ # 43 ]

Just saw this https://www.ted.com/talks/oscar_schwartz_can_a_computer_write_poetry?language=en

Love the reverse Turing Test lol

 

 
  [ # 44 ]

AIML can also be used for poetry creation. Say to Mitsuku, “Tell me a poem about x” where x is a subject such as cakes, dog, river or whatever. Ask her again and she should create a different poem. Not bad for a pattern matcher wink

 

 
  [ # 45 ]

I was lazy and did not read through all the posts. My only real comment is what is intelligence? How do you define it? What makes one species more intelligent over another? Intelligence is relative to the environment. To cat a human would be stupid because we can not hear or see all the same things they can. To a bacteria we would not even exist as it is beyond its ability to comprehend but yet it does things without a brain. Are machines intelligent and can we really say without some self imposed Narcissism say that we are some how more intelligent then them?

 

 < 1 2 3 4 > 
3 of 4
 
  login or register to react