People use I fuzzy pattern match. The following example demonstrates how most people can still read the text even though only the first an last letter are correct.

__

I cnduo’t bvleiee taht I culod aulaclty uesdtannrd waht I was rdnaieg. Unisg the icndeblire pweor of the hmuan mnid, aocdcrnig to rseecrah at Cmabrigde Uinervtisy, it dseno’t mttaer in waht oderr the lterets in a wrod are, the olny irpoamtnt tihng is taht the frsit and lsat ltteer be in the rhgit pclae. The rset can be a taotl mses and you can sitll raed it whoutit a pboerlm. Tihs is bucseae the huamn mnid deos not raed ervey ltteer by istlef, but the wrod as a wlohe. Aaznmig, huh? Yaeh and I awlyas tghhuot slelinpg was ipmorantt! See if yuor fdreins can raed tihs too.

Yes, though that is reading, which is a visual process, also pattern matching though.

Note how the avatar of the woman keeps moving its lips after her audio stops at about 0:37, at the end of the statement “But you said earlier that you were a robot.” And note how the video implies that they’re hearing each other (doing speech recognition), which makes it much more interesting (I bet) to the casual viewer. But in fact the bots are processing text, like Watson was sent the text version of the Jeopardy questions (or answers if you prefer) behind the scenes…

Re the Cambridge study, this page http://www.languagehat.com/archives/000840.php cites the counterexample:

Hilpapy aoutrhs sitll issnit on fwinollog dullfready oslotebe snellpig.

Also what about words that have the same first and last letter but different letters in between, like “tehre”. Again from the link:

Let’s see if this pertains to painters of pantries done in loco parentis.

—-

An anagram descrambler could be used to decipher words, such as mine at http://subbot.org/descrambleragent/ :

> Aoccdrnig
according

> rseearch
reachers research searcher

> iprmoatnt
important

> cnduo’t
Default response.

> bvleiee
believe

> pretains
painters pantries pertains pinaster pristane repaints

> dullfready
dreadfully

In cases where more than one dictionary word matches the scramble, and where the first and last letters are the same, you would have to do word sense disambiguation of course. (This is probably fairly rare, as in the “painters” example; “three” and “there” might occur more often, however.) Also note my program doesn’t handle contractions but that doesn’t seem too hard to add. And my program didn’t get “parentis” so I need a way to teach it new words (also should be easy to add to the dictionary).

If anyone else wants to write a program that can descramble text like the Cambridge study example, I bet I could write one that would do it faster and more accurately! :)

That should be, “what about words that have the same letters but the ones between the first and last occur in a different order, as in ‘there’ and ‘three’.”

One of the comments in the language hat link (http://www.languagehat.com/archives/000840.php) referred me to http://www.ai-forum.org/topic.asp?forum_id=1&topic_id=9528&page=1 where Ted Warring posted a link to his (now 8-year-old) program http://www.ai-forum.org/data/30-fr_small.zip which deciphers the scrambled text:

I CNDUO T believe that I cloud actually understand what I was reading using the incredible power of the human mind according to research at Cambridge university it DSENO T matter in what order the letters in A word are the only important thing is that the first and last letter be in the right place the rest can be A total mess and you can still read it without A problem this is because the human mind does not read every letter by itself but the word as A whole amazing huh yeah and I always thought spelling was important see if your friends can read this too

Note that “couldn’t” is misspelled in the original; the program misinterprets “could” as “cloud” (a word sense disambiguation problem); and the program doesn’t get the contraction “DSENO’T”.

—-

http://www.mrc-cbu.cam.ac.uk/people/matt.davis/cmabridge/index.html has the example sentence:

“The sprehas had ponits and patles”

Might come out as…

The sherpas had pitons and plates.

The shapers had points and pleats.

The seraphs had pintos and petals.

The sphaers had pinots and palets.

The sphears had potins and peltas.

Warring’s program outputs: the shapers had points and petals.

(I can add a rule at runtime to my descrambler to eliminate solutions that don’t keep the first and last letters:

> descramblerbot: if input =~ /^(.).*?(.)$/ then first = $1; last = $2; responses = response.split(’ ‘); toReturn = Array.new; responses.each { |r|; if r !~ /^#{first}.*#{last}$/ then next end; toReturn.push(r); }; response = toReturn.join(’ ‘); score = 5 end
Okay [...]

> sprehas
seraphs shapers sherpas
[I would have to add sphaers, an old form of “spheres”, to the dictionary…]

> pretians
painters pantries pertains)

—-

http://www.livejournal.com/users/kinkyturtle/87224.html has the passage:

“Artichoking to a reindeerwash at Cartilage Unfamiliarity, it dentis’t molar in wheat oater the ladybugs in a weed ate, the ozzy insideout thong is twit the foist and list louvre be at the ranft pekoe. The raft can be a Tyrol MIDI and you can shell rind it whippet pogrom. Toes is buncombe the harem mood digs not road energy ladder by insectivore, but the wild as a Wookiee.”

Dave Morton - Sep 1, 2011:

To be brutally honest, Erwin, practically every chatbot now in use (that I know of) cannot really be considered to be “AI”, since they rely on “basic pattern matching” of some sort or another, rather than any generative sort of algorithm to produce an output. To be sure, some chatbots are closer than others to the concept of AI, but none that I know of are there yet. But as they say, “it’s the journey, not the destination, that’s important”, so I say, “ROAD TRIP!!!”

I agree, Dave.

I question whether or not true Artificial Intelligence is achievable. I’m of the opinion that AI is an oxymoron, and not merely as a joke. Even Watson relied on scanning articles related to keywords for answers to Jeopardy questions, while our pets have been shown to display creativity and to exhibit an ability to deduce an outcome.

Is mere memory intelligence? I think that a large storage potential along with a rapid retrieval system is something less than intelligence. I have a cat that does some pretty amazing things and, I believe, shows genuine intelligence. Maybe Artificial Intelligence is actually a misnomer and should be labeled as something else because we can’t help but measure it against human intelligence.

Whatever people settle on calling it, it’s never been more than emulating intelligence, or giving the perception of intelligence. People with pets or children understand what real intelligence is, and we’ve observed the process of “thinking” in biological entities. Computers and programs don’t “think” they “search” and copy.

If there is some form of AI out there, and I can talk to it without having to go several floors beneath the Pentagon, I’d like someone to point me to it.

Thunder Walk - Sep 3, 2011:

If there is some form of AI out there, and I can talk to it without having to go several floors beneath the Pentagon, I’d like someone to point me to it.

Waits for Arthur to pop his head up…

Thunder Walk - Sep 3, 2011:

If there is some form of AI out there, and I can talk to it without having to go several floors beneath the Pentagon, I’d like someone to point me to it.

There is embedded A.I. in industrial computers all around us, which we may take for granted, but some specialist still has to turn the knowledge of a human expert, into computer code for the A.I. to be embedded into the medical device.

For example, getting an EKG is common with “interpretation” A.I. embedded on a chip in the medical device, which in a way, “talks” to the doctor about what the patient’s heart said to it.  In this case, machine interpretation saves lives.

The alternative is to have a human expert read the EKG printout to manually “interpret” a diagnosis.  There are other examples of industrial robots being used in hospitals, to dispense medications, but perhaps none are as serious as fighting heart disease, our number one killer.

Pharmaceutical dispensaries have been monitoring a patent’s drug prescriptions looking for potential conflicts for years. I think that’s a good time-saving and labor-saving use of computers, but I don’t view that as artificial intelligence.

Various military training schools have a method of presenting a group with a problem, along with several tools, none of which alone is sufficient for surmounting the obstacle. The object is to solve the problem creatively.

When I think of artificial intelligence, that’s the kind of thing that comes to mind.

Thunder Walk - Sep 3, 2011:

[...]If there is some form of AI out there, and I can talk to it without having to go several floors beneath the Pentagon, I’d like someone to point me to it.

Taking Steve Worswick’s bait, I give you

http://www.scn.org/~mentifex/AiMind.html

as an artificial intelligence you can talk to. At the same time, here is a progress report on the state of Mentifex AI. I do not often post here, because I have been extremely busy coding

http://www.scn.org/~mentifex/mindforth.txt

which is the more powerful AI Mind for which the JavaScript AiMind.html serves as a tutorial—one that will think just as well as MindForth until the two programs diverge at some point in the future. It is somehow easier to code the AI Mind in Forth than in JavaScript. Then it is rather easy to port the improvements from Forth to JavaScript.

A lot has changed in these Mentifex AI Minds over the summer of 2011. Many peripheral mind-modules have been fully fleshed out. It may turn out to be controversial, but I have been coding some “DeFault” settings into various mind-modules. For instance, if the WhatAuxSVerb module is supposed to ask, “What do birds eat?” but the verb “eat” has gotten lost, the module will default to “What do birds do?” It’s cheating, but the AI programmer notices it and can take corrective action. I am also gearing up to code a “DeKi” (deutsche künstliche Intelligenz) or German-language AI, just because I love the language of Beethoven., Heine and Nietzsche. German has so many irregular verb and noun forms, that I will put a bare-bones DeKi up on the Web and then try to “crowdsource” the future development of German AI.

About twenty entities a day visit the Mentifex JavaScript AiMind.html, and a lot fewer visit the MindForth AI, but the interest from Netizens keeps me strongly motivated to keep on programming the artificial Minds.

One further thing about “crowdsourcing”. In the MindForth AI there is a lot of white-space, especially with IF-THEN clauses, so I have begun filling in the lines that have only “THEN” as a word of code and I have been commenting-in the URL addresses of obscure AI projects that I find on the Web, because I want to encourage AI Trek The Next Generation. -Arthur

Here’s what I think I know…

The term Artificial Intelligence comes in different flavours - weak, strong etc.. and you could also have subsets I guess. For some reason agreeing on what AI means is a difficult problem wherever I go.  Everyone has their own opinion.

I’ve seen the words ‘True AI’ bandied around before and it’s hard to know what that is supposed to mean really. So we make AI and then we make true ai ??  Is that the idea ? So AI in itself is nothing much ?

All very subjective a lot of the time. Not necessarily a bad thing by the way.

The way I see it is that intelligence varies a great deal in nature. Sure we are smarter than rats and maybe dolphins are smarter than us. A bee on it’s own is pretty smart, it knows what to do to survive - when they are in a hive or swarm they become something else entirely. It’s all intelligence to me.

And so it follows, for me at least, that you get varying degrees of ‘intelligence’ in machines too - and so we get ‘Artificial Intelligence’ - a wide and varied array of creations, some smarted than others.

I think maybe True AI to some people means sentience, but I don’t think that is going to happen in my lifetime if ever.

That’s my 2 pennies anyway

Fred… I mean, Roger, that has been the topic of several discussions, not only here, but just about anywhere that two or more people involved in AI gather to discuss the issues of the day.

I’ve been following (and involved in, as well) a similar discussion (I hope that link works) over at LinkedIn that has proven to be quite enlightening, as I’m finding the people there to be just as passionate about the subject as many of us seem to be, here. I’m finding that, both here, and elsewhere, my notions and ideas about AI are constantly being challenged, and I’m exploring new ideas practically every day.

I agree with you regarding the notion that some people equate intelligence with sentience. The two concepts are worlds apart, IMHO, but not everyone agrees with my ideas. Not the foggiest idea why, though.

I agree that there are different opinions (subtle differences, really) with regard to the definition of Artificial Intelligence.  Still, there are also some obvious similarities that are impossible to ignore, and which we can agree upon.

I believe that, at this level (meaning something less than rooms-full of banks of Cray computers where teams of scientists from a variety of fields toil over keyboards, experimenting with theories we haven’t even thought of) bots can “emulate” conversation based on combinations of keywords.  It’s not complicated, it’s not difficult, and it’s not top secret mumbo-jumbo, magic, or anything all that amazing.

However, on a regular basis, we can observe birds, fish, animals, and even insects demonstrating something we recognize as “intelligence.”  To list would take up space and waste time, most know what I’m talking about.

In the example I gave above with regard to, “military training schools,” the mere parsing of language won’t “solve” problems.  While it might be a component, and assist the program/computer in approaching a problem, it won’t answer, “There are numerous people of an unknown affiliation who have taken bus-fulls of soldiers hostage all across the country, and are threatening hostile action in three hours if we don’t comply with their demands to turn the government over to their political organization, what is the best plan of action?”

On the other hand, if all you care about is that a bot somehow comprehend that cats chase birds, and that’s your definition of AI, then your belief is secure.

Interesting comments guys, I know what you are saying I think.

Thunder Walk - Sep 5, 2011:

On the other hand, if all you care about is that a bot somehow comprehend that cats chase birds, and that’s your definition of AI, then your belief is secure.

Yes I think that is true really.  I think it’s just a question of purpose.  That’s all some AI’s will need to do…

Just because it can’t tell us the meaning of life it doesn’t necessarily mean it’s useless.

My point is really that all this stuff is AI, however advanced or not it is.  It’s a generic term. All that leaves us to do is quibble over what one is better than the other and find the ones that are useful to us. Which we all love to do ;P

Edit, thanks Dave M, I will take a look at that link.