Language and intelligence
« previous post | next post »
Two interesting popular articles on linguistic aspects of artificial intelligence have recently appeared in the popular press.
The first one is by Richard Powers ("What is Artifical Intelligence?", NYT 2/6/2011):
IN the category “What Do You Know?”, for $1 million: This four-year-old upstart the size of a small R.V. has digested 200 million pages of data about everything in existence and it means to give a couple of the world’s quickest humans a run for their money at their own game.
The question: What is Watson?
I.B.M.’s groundbreaking question-answering system, running on roughly 2,500 parallel processor cores, each able to perform up to 33 billion operations a second, is playing a pair of “Jeopardy!” matches against the show’s top two living players, to be aired on Feb. 14, 15 and 16.
Powers' novels include Galatea 2.2 (1995), in which a fictional Powers
… meets a computer scientist named Philip Lentz. Intrigued by Lentz's overbearing personality and unorthodox theories, Powers eventually agrees to participate in an experiment involving artificial intelligence. Lentz bets his fellow scientists that he can build a computer that can produce an analysis of a literary text that is indistinguishable from one produced by a human. It is Powers' task to "teach" the machine. After going through several unsuccessful versions, Powers and Lentz produce a computer model (dubbed "Helen") that is able to communicate like a human. It is not clear to the reader or to Powers whether she is simulating human thought, or whether she is actually experiencing it. Powers tutors the computer, first by reading it canonical works of literature, then current events, and eventually telling it the story of his own life, in the process developing a complicated relationship with the machine.
The other article on language, humans, and machines is in the most recent Atlantic, where Brian Christian describes and discusses his experiences as a participant in the 2009 Loebner Prize competition ("Mind vs. Machine"):
I wake up in a hotel room 5,000 miles from my home in Seattle. After breakfast, I step out into the salty air and walk the coastline of the country that invented my language, though I find I can’t understand a good portion of the signs I pass on my way—LET AGREED, one says, prominently, in large print, and it means nothing to me.
I pause, and stare dumbly at the sea for a moment, parsing and reparsing the sign. Normally these kinds of linguistic curiosities and cultural gaps intrigue me; today, though, they are mostly a cause for concern. In two hours, I will sit down at a computer and have a series of five-minute instant-message chats with several strangers. At the other end of these chats will be a psychologist, a linguist, a computer scientist, and the host of a popular British technology show. Together they form a judging panel, evaluating my ability to do one of the strangest things I’ve ever been asked to do.
I must convince them that I’m human.
Fortunately, I am human; unfortunately, it’s not clear how much that will help.
Christian's special take on the competition is that he doesn't really care about how the computers do — he wants to be able to prove that he's human; or rather, he's interested in what he has to do to demonstrate his humanity in a conversation, and he gives entertaining examples from previous competitions that focus on this question.
In the end, he does win the "Most Human Human" award, as the human entrant with the best score in a round-robin against the computer entrants.
In some ways a closer fight would have been more dramatic. Between us, we confederates hadn’t permitted a single vote to go the machines’ way. Whereas 2008 was a nail-biter, 2009 was a rout. We think of science as an unhaltable, indefatigable advance. But in the context of the Turing Test, humans—dynamic as ever—don’t allow for that kind of narrative. We don’t provide the kind of benchmark that sits still.
As for the prospects of AI, some people imagine the future of computing as a kind of heaven. Rallying behind an idea called “The Singularity,” people like Ray Kurzweil (in The Singularity Is Near) and his cohort of believers envision a moment when we make smarter- than-us machines, which make machines smarter than themselves, and so on, and the whole thing accelerates exponentially toward a massive ultra-intelligence that we can barely fathom. Such a time will become, in their view, a kind of a techno-Rapture, in which humans can upload their consciousness onto the Internet and get assumed—if not bodily, than at least mentally—into an eternal, imperishable afterlife in the world of electricity.
Others imagine the future of computing as a kind of hell. Machines black out the sun, level our cities, seal us in hyperbaric chambers, and siphon our body heat forever.
I’m no futurist, but I suppose if anything, I prefer to think of the long-term future of AI as a kind of purgatory: a place where the flawed but good-hearted go to be purified—and tested—and come out better on the other side.
The scores and transcripts for the 2009 Loebner competition are here.
Christian's article is adapted from a book to be released on March 1, The most human human: What talking with computers teaches us about what it means to be alive.
Some previous LL posts on relevant topics:
"Philologists say…", 9/28/2005
"Thriving on confusion in the Guardian", 5/24/2006
"To what after shampooing?", 10/20/2007
"For any transformation which is sufficiently diversified…", 7/4/2008
"The Dowdbot challenge", 5/21/2009
"Our love was real", 9/7/2009
"Botty man", 9/14/2009
"Richard Powers on his way to a decision", 10/28/2009
"Just add 'intelligent' and 'informed'", 10/26/2010
"The robot army", 11/28/2010
"The case of the missing spamularity", 12/23/2010
I would have sworn that I posted something about this video when it came out early last summer, but apparently not.
Twitter Trackbacks for Language Log » Language and intelligence [upenn.edu] on Topsy.com said,
February 9, 2011 @ 7:07 am
[…] Language Log » Language and intelligence languagelog.ldc.upenn.edu/nll/?p=2952 – view page – cached February 9, 2011 @ 6:32 am · Filed by Mark Liberman under Computational Linguistics, Language and the media Tags […]
LING 255 » Blog Archive » Some fun reports from the world of Turing Tests said,
February 9, 2011 @ 9:07 am
[…] This is somewhat tangential to our course, but if you're curious about artificial intelligence, you might enjoy this post on Language Log (and the articles it points to): Language and Intelligence […]
language hat said,
February 9, 2011 @ 9:12 am
I went through the second half of that post waiting for you to tell me what "LET AGREED" meant. Since I received no satisfaction, I will save others the effort by providing this definition (courtesy of Yahoo Answers):
[(myl) Sorry, Hat, this comment was caught for a while in the spam filter. I apologize on behalf of Akismet.
I found the same Yahoo Answers definition that you did, and was thinking about putting it at the end of the post, but the coffee-maker beeped and then I forgot. Thanks for adding this.]
Martin J Ball said,
February 9, 2011 @ 9:15 am
Let Agreed??
I've never seen a sign with that wording in Britain…?
Might it have been "Give Way" (='yield')? Maybe his memory got degraded :)
[(myl) And yet …]
John Cowan said,
February 9, 2011 @ 10:09 am
Hat: The OED2 defines let n. 2 thus:
A letting for hire or rent. (The sense in the first quot. is doubtful.)
1684 in A. Nora Royds Reg. Par. Felkirk (1896) 3 By ye Ancyant Lett it amounts to 35 Pounds Yearly.
1839 Dickens Nicholas Nickleby xxiv. 227 ‘We've had a pretty good Let,’ said Mr. Crummles. ‘Four front places in the centre, and the whole of the stage-box.’
1868 Perth. Jrnl. 18 June, John Dewar, at the Farm, will show the Boundaries; and the Conditions of Let may be learned on application.
1878 Daily News 24 Oct. 6/6 The reason the stair was not included in the lease was that the executors wanted to utilise it for the empty rooms, and make a separate let of it.
Spell Me Jeff said,
February 9, 2011 @ 10:34 am
There is something wonderfully Orwellian/Philip K. Dickian about having to prove you're human.
I taught ESL to undocumented aliens during the 80's amnesty program. To be eligible for amnesty, you had to prove that you had lived in the country for X number of years. Yet most immigrants had worked very hard to keep their existence unknowable. Such evidence was therefore hard to produce.
Zythophile said,
February 9, 2011 @ 10:55 am
@Martin J Ball – you need to get out more.
richard howland-bolton said,
February 9, 2011 @ 11:15 am
Ah! I remember, as a 'youf' back in England, that the normal sign on rental property was "TO LET"; into the word-space of which some wag invariably inserted an 'I'.
stormboy said,
February 9, 2011 @ 11:52 am
So what's the US equivalent of 'Let Agreed'?
[(myl) "Rented". I think.]
Not My Leg said,
February 9, 2011 @ 12:40 pm
I think "rented" is correct. 'Let agreed' and 'to let' are both generally alien phrases in the US. We would almost always see 'for rent' and 'rented' or perhaps 'for lease'.
Steve the Steam Shovel said,
February 9, 2011 @ 12:44 pm
Ah! I remember, as a 'youf' back in England, that the normal sign on rental property was "TO LET"; into the word-space of which some wag invariably inserted an 'I'.
Which is why the normal sign was more usually "TO-LET", even if the hyphen names no sense.
J. W. Brewer said,
February 9, 2011 @ 1:18 pm
30 years ago the shelves of my U.S. public high school library contained a novel entitled "TO LET" (wikipedia makes me think perhaps the Galsworthy one concluding the Forsyte saga?), into which the obvious "I" had been anonymously inserted on the spine. The "FOR RENT" meaning of the original title was probably completely opaque to us.
Keith said,
February 9, 2011 @ 2:23 pm
I understood "Let Agreed" immediately, as being a sign outside a house; the sign has replaced another that read "To Let".
I am sure that I have seen the terms "lessor" and "lessee" in the USA, since I moved here, and a quick Google search finds them in American law articles.
K.
Chandra said,
February 9, 2011 @ 2:57 pm
I figured out "LET AGREED" too, but I still think it's weird. I can see "let" as a verb meaning "rent" in the phrase "to let", but wouldn't "lease agreed" make more sense than "let"? Is the rental agreement, or the property itself, known as "a let" in other phrasings?
KevinM said,
February 9, 2011 @ 3:07 pm
I suppose if two tennis referees came to an understanding as to the proper call…
J. W. Brewer said,
February 9, 2011 @ 4:17 pm
I wonder if "LET AGREED" means more or less what "in contract" means in U.S. residential real estate transactions, which is in effect: "we have a signed deal, the property is currently not being shown to anyone else, but the deal hasn't actually closed, the new people haven't actually moved in, and I the broker certainly haven't gotten my commission yet; so if you're interested in the property, keep checking back in just in case the current deal falls apart and the property goes back on the market." That would be a reason to put a new sign up, rather than just take the "TO LET" sign down and leave the now-leased premises unmarked.
Ken Brown said,
February 9, 2011 @ 8:43 pm
@chandra, yes, "a let" works. Mostly in fixed phrases lilke "a short let". (which gets loads of Google hits)
army1987 said,
February 10, 2011 @ 6:32 am
S***w you richard howland-bolton, I was just about to write about that!