Slashdot Mirror
Is Google's Future: Star Trek?
An anonymous reader writes "ZDNet UK has an interview with Google's CTO, Craig Silverstein, and he's got some pretty cool visions: "When search grows up, it will look like Star Trek: you talk into the air ("Computer! What's the situation down on the planet?") and the computer processes your question, figures out its context, figures out what response you're looking for, searches a giant database in who-knows-how-many languages, translates/analyses/summarises all the results, and presents them back to you in a pleasant voice." Now that's the search engine I want." The NLP required for this is far off, but it sure will be cool when we get there.
Make it so. :)
Mod me down with all of your hatred and your journey towards the dark side will be complete!
I know google's great and all, but this is basically a "we want to be able to do everything cool with computers and AI, but we don't know when that's going to happen" type story.
I imagine if you ask Microsoft, Apple, or Palm, they'll mimic those goals. NLP, instant searching, instant translations, it's all well and good, but where's the story?
--
Use Vobbo for Video Blogs
Captain: Tea, hot, Earl grey.
Computer: Did you mean Hot Teen URL's
This week, not only will we have answered the question of just how much of our knowledge we base from the Internet (Google, by and large), but how we can make it even easier to use. Anyone see any searchable database on the Web with the potential to topple what Google has become / could become?
I know nothing
[Scotty talking into Mac+ mouse] Computer? Hello computer?
Must-not-watch TV!
...does it *have* to have the voice of Majel Barrett?
You must think in Russian.
With technologies such as quantum computing down the road, I couldn't possibly envision a future where this isn't a possibility.
There was a short on NPR that explained it the best: Imagine looking for a person when only knowing their phone number. Today we look through the phonebook one name at a time, but with quantum computing, we'd look at the entire phonebook at once.
Are not that far off, but with my work in AI or "smart" anything devices, always come up with the same results. The weakest link is not the technoligy but more so the people using it. Remember, everytime you think you make something idiot proof, they build a better idiot.
"It's better to be a pirate then join the Navy"
It's about voice recognition and its reliability. I think that everyone expects that this future is inevitable but, until voice recognition reaches a point were it can reliably interpret a vast vocabulary from multiple voices and accents, none of this can happen.
To be sure, progress is definitely being made in voice recognition technology. But, that progress is slow and we are still many stardates away from success.
As humans a lot of our brainpower is geared towards interpreting visual input. Its will always be a lot faster for me to look at the pages of hits returned and determine what is of interest to me than it will be to listen to a computer voice and try to figure it out. Speaking to the computer is OK but in many situations I will want visual, not aural feedback
Do not try to read the dupe, thats impossible. Instead, only try to realize the truth
What truth?
There is no dupe
You will ask "Computer! What's the situation down on the planet?" and you get 100 sites, all linked to each other, that have this phrase crammed into a mass of links and search-engine-bait, all trying to sell you cable de-scramblers and viagra.
Ever notice the 'rot' that is occuring on google lately? For example, a search on "mercedes 300D transmission" used to bring up the article on mbz.org about adjusting the vacuum shift in this car. Now this link, the most useful one, is all the way on the third or fourth page, buried in OEM parts retaillers that you know damn well are ranked high thanks to "ranking services".
I hope they can figure out how to weed this kind of stuff out...
+++ ATH0 +++
Disclaimer: I did write one of the papers.
...so this'll be available shortly after the transporter?
:-)
It's a Nice Thing, but does anyone have any insight into exactly how far off into the future we are looking?
- speech recognition systems leave quite a lot to be desired
- is there *anything* out there that's able to put stuff into context {so to speak}
- if it's far enough off, the whole multiple-language thing will take care of itself - the number of languages is dwindling each year
will be nice to have, though.
Oh - wait. Probably won't be in my time. And if it is, I'll probably be to old to figure out how to use it. Drat all these new-fangled things!
.sig? No.
The actual [SCO] lawsuit is very narrow in its claims; we're not nervous about it at all. It's prompted lots of discussion, which has been very interesting to watch.
:)
Somebody reads slashdot
On webmasterworld this very topic is discussed all the time (though mostly by search engine optimizers who apparently have nothing better to do with their time). If you can put up with the marketroids, it's actually a very useful website.
Alltheweb and Teoma seem to be Google's most credible challengers technology-wise, although Microsoft is also now developing its own search engine.
Google, seeing the risk, overhauled their search engine this summer--I wonder if anyone here has noticed the difference.
Yeah, because there's all those websites that talk about faucet and washer porn without using the word how.
Except, when I searched that phrase, the first link is "How to fix leaky faucet," then "Fred and Gerry on leaky faucets," another, "How to fix leaky faucet," next is "Repair a leaky faucet in six steps," then "Repair a leaky washer-type faucet," and it just goes on from there.
Too bad none of those had to do with fixing leaky faucets.
Natural Language Processing or voice recognition.
I guess that there are still those amongst us that insist on trying to supplement their inadequacies by babbling in acronyms.
I've always said that if you think it's cool or leet to speak using acronyms, you should go all out and speak in hieroglyphics.
Funny, the first link returned is a great explanation of how to fix a leaky faucet. Want to try again?
More context for that quote:
"When search grows up, it will look like Star Trek: you talk into the air ("Computer! What's the situation down on the planet?") and the computer processes your question, figures out its context, figures out what response you're looking for, searches a giant database in who-knows-how-many languages, translates/analyses/summarises all the results, and presents them back to you in a pleasant voice. I think this technology is about, oh, 300 years off. Just getting the computer to understand your question, much less the context it's being asked in, is way beyond the state of the art in computer science right now."
There are no trails. There are no trees out here.
Then again, I would also very much like to see flying cars, true AI, useful nanotechnology, and practical fusion reactors.
All of which, of course, are things that we've been promised for years/decades are "right around the corner", yet always fail to materialize.
They were saying we'd all be in flying cars now in the 1960s... where's my aero-Ford?
Honey, I shrunk the Cygwin
That, and I don't really want my coworkers hearing "Computer, get me some boobies!" from my office.
You know what?
You can't perform speech processing with a computer until we have a computer that is built like a neural network which understands speech as we do. As a side effect such a computer wouldn't have to perform serialized "searches"; once you asked it a question it would already know the answer.
The future of searching is: Computers will NOT search as they do today. They will be based on the model of the human brain and how it addresses "memory", by activating nodes in a massive neural network.
Natural Language Processing or voice recognition
Actually the two are distinct but related concepts...
Natural Language Processing is the science of how to take a grammatical statement and parse it. Breaking it down into nouns and verbs and subjects and objects and whatever, and then representing the symantic links that describe how these concepts modify each other in a grammatical context.
Voice recognition is the science of taking spoken language and transcribing it to a context-specific computer representation.
The two technologies can be married, in that the context-specific output of voice recognition be a NLP parsing structure... but they don't have to be. Back when I was reading AI mags every week (about 4-5 years ago), all the voice recognition guys were outputing ASCII and all the NLP guys were inputing ASCII but that was as close as they got to working together...
I am disrespectful to dirt! Can you see that I am serious?!
Visual is quicker than audio, and one of the reasons is that it is nonlinear (not in the mathematical sense, but the dimensional sense).
Visually you can look at a screen of replies and skip to the next "line" instantly if the current line is not what you want. Difficult for audio.
Also, you can look at the screen as a whole and can often see the answer you desire because it essentially "jumps out at you" from visual filtering. Listening to all the audio output as a whole will most likely give you nothing.
So what would really be cool (and more practical) is to have the voice input (a la Star Trek) and then have an instant display of results on a viewable surface in mid-air.
> "Computer! What's the situation down on the planet?"
"How does it make you feel to ask what's the situation down on the planet?"
- For the complete works of Shakespeare: cat
...thought of in the 80s when they created the Knowledge Navigator clip. Scully's dream was to eventually create a computer that would act as an assistent that you could also ask questions. It would come back later when it found answers. Of course, the whole concept was a pipedream, but still, the Newton's 'Assist' button was one of the first steps towards that goal.
Too bad Jobs had to kill the Newton when he got back at Apple to finally do away with everything Scully.
Voyager, Series 3, Distant Origin
"Computer, display the likely appearance of this creature given 300 million years of evolution"
How long did it take to get a result?
About half a second.
"Taligent is still pure vapor. Maybe they'll be the last who jumps up on Openstep... "
Voice Recognition is essential in hospitals and courts all over the US.
It is a multi-million dollar a year buisness.
As for the future of speech recognition, people often make the mistake of seeing it as a replacement for all other input interfaces. This is just absurd, as you point out.
More and more products (like cell phones, car GPS, ATM's, LARAN ) use speech recognition in conjunction with other more traditional means of input/output to get the job done.
It is NOT the fastest or the most reliable means of interfacing, but it is the most NATURAL.
You're kidding, right? Do you even remember how searching was before Google came around? Google revolutionized Internet searching, and last time I checked they continue to lead the pack. They get a lot of publicity because a lot of people look to them for the next big thing (and rightly so IMHO).
-- Kircle
It's an average search engine !
Google is an average search engine? Let me guess, you started getting downloaded on the internet sometime around 1999.
You don't remember Alta Vista, Yahoo, or the countless others before Google. I switched to Google exclusively when it was still in beta.
Nothing unique in their software.
There is something unique, it's called PageRank. You may have seen it in the freaking patent system.
Apparently "Interesting" is now a synonym for "Factually Incorrect"
Dacels Jewelers can't be trusted.
Google ought to do complex Boolean queries like
(potato or potatoe) and ((fried or mashed) and gravy)
It's my only peeve about that wonderful search engine.
That that is is that that that that is not is not.
Where every team has to have one employee whose sole job is to talk to the computer!
"Freedom means freedom for everybody" -- Dick Cheney
Considering Data wrote new subroutines for himself fairly frequently I never understood why couldn't cope with s/will not/won't/g etc.
That would've required a kernel patch, and he didn't want to ruin his uptime.
Hell it even tells you the life, universe and everything!. + + + + Only thing I noticed, google images doesn't cache the goatseman's pic... :(
"I am slashbot, hear me roar!"
What you just said is the "geek perspective" on things, and I agree with that.
However, do NOT underestimate the desire that "average Joes" have to be free of keyboards.
When you're a blue-collar factory worker with a dumpy crappy Compaq running Windows XP Home and connecting to the Internet through AOL, and you can type all of 5 words per minute on a GOOD day by hunting and pecking, the one thing you want the MOST is to be able to talk to the thing. I predict that in the future, keyboards will ONLY be used by programmers (as we're virtually the only ones who need to type funky things like "printf("Hello, World\n");" that would be a RIDICULOUS pain to input with voice), and they will cost a huge amount. Also, it is likely that they will only work with Windows. KEYBOARDS ARE GOING TO BECOME EXTREMELY RARE, and hence EXTREMELY EXPENSIVE, since most people CAN'T type faster than they can talk.
Honey, I shrunk the Cygwin
Microsoft already spends billions on R&D every year and even has an existing state of the art facility on their Redmond campus.
Nothing good has come of it in all the years of its existence.
Mac OS X and Windows XP working side by side to fight back the night.
I haven't seen any "WOW!" things come out of the project yet, but you have to admire their "just do it" approach to AI.
Isn't this what AskJeeves strived to do from the beginning?
Interpret your question and hopefully give you a suitable answer... but it's not perfect yet.
Come to think of it, isn't that also what Clippit/Clippy tried to do, much to the world's chagrin?
Google is on the verge of sucking. It USED to take me to what it want. Now, it tends to take me to where it wants to go.
Linux search terms tend to take me to the wrong places all the time. (Google groups works better). Enternainment and move title search times take me to the wrong places. Generic searchs "whats the weather like in Mountain View california" are AWFUL. Searching for hotel/resort information in an area is awful (takes me to package tour sites).
In all these examples, Google takes me to "commercial clearinghouses" rather than the definitive source of information. The more successful Google is whenever it IPOs, the more its results will skew. (Google groups searchs still tend to take me "where I want to go").
All of this reinforces my theory: all search engines are doomed to fail. The start small, peak, get rich, suck, and go away. Google is entering the "get rich" phase.
As I was reading the comments attached to this story, one point kept coming to mind. Maybe, just maybe, this type of idea could be the savior of human language.
Bear with me on this.
Leet-speak aside, vocal (as well as written) communication has (IMHO) been deteriorating at a rather rapid pace. Now, it could just be the fact that I am working in a direct customer contact position again, and I have to deal with the general public on a more frequent level than I used to. But it simply amazes me the number of people who cannot communicate what it is that they are thinking.
"I am looking for one of those orangishy whatchamacallits wit' that springy thingish-like doohickey on the end"
He wanted a pipe wrench.
*eep*
If this technology were to become as ubiquitous as google has become as a search engine, people who wanted to be able to use this technology would have to learn to communicate clearly and concisely.
(Yes, I am well aware of the fact that I have certainly not mastered those skills myself, so please don't flame me, its just an idea)
In reality, the ACLU would probably sue the programmers until the language heuristics were so loose that it would become unusable, because some idiot with money and power got upset because his new "Google enviromental information interface" kept telling him that there was no such word as nookular.
Wouldn't it be great though, to see people actually interested in learning how to communicate better, because they have been given a technological incentive, instead of dumbing down the interface because they are too lazy to learn how to use it?
</Pipe Dream>
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order- Ed Howdershelt Via Tass
Please visit the web site for the natural-language processing (NLP) laboratory at Microsoft Research.
That's part of it. But the bigger problem I see with this scenario is getting humans to verbalize what they're really looking for. I work for a public library, answering computer questions for the public. Finding the answer is not the biggest problem. The biggest problem is getting the public to accurately explain what the hell they're looking for.
That requires two things:
1. Knowing what they really are looking for
2. Being able to verbalize it
In some ways, the written word is superior because often when they write the actual words, people are more specific about what they need. Usually they've considered it and narrowed it down a bit (though not always).
Real life examples of humans searching for info:
"Where are the art books?" Actual need: tattoo information
"I need a book on Microsoft." Actual need: Learning that the Enter key will move you down to the next line when using a word processing program such as Word
"When I was little, I really liked this book you had. The little girl in it was named Jane or Joan, I think. I think it was blue. Do you know it?"
As you can see, many people do not give enough information or context on their first try. So computers would have to learn how to ask questions for more input and get people to narrow things down. And while that's easy in some situations, it can be difficult to guess the correct context in others.
That technology seems years away to me.
Acronym Finder.com has several relevant solutions:
Name Lookup Protocol
Natural Language Processing
Network-Layer Packet
Neuro-Linguistic Programming
Non-Linear Processor
Nonlinear Programming
I found the use of it sophomoric myself.
nothing unique??
What about Pigeon Rank??
http://www.google.com/technology/pigeonrank.html
Fire in the hands of the village idiot is no tool, but a weapon of mass destruction
Go to Cnet and the download area. You will see the pages ranked by most requested downloads. And anyways Yahoo uses most requested links in their Categorized Links on their home page. Such as Computer>Supercomputers>Cray. While this was categorized by humans it still just as good as good as what Google does.
Hey kid, before you embarass yourself any further why don't you just go ahead and look at what PageRank actually is. Google did invent what they call "PageRank." It's name is a touch misleading.
Dacels Jewelers can't be trusted.
There's a simple reason for this:
:) )
It's quite easy to build a system that analyses something for a certain property, be it the net, the stockmarket, society, etc. etc. unfortunately, as soon as this system becomes well known, everyone tries to manipulate it. In the stockmarket people try to create formations common in technical analysis to make other traders buy/sell a certain stock, and in internet searching people set up huge arrays of pages referencing each other or scatter ridiculous numbers of irrelevant key-words over their page.
I think i read it the first time in one of the old Asimov books, that to predict something well the predicted system must have no knowledge of the prediction... (note to physicsts, i'm talking of systems involving people, not a mass on a spring
Ponxx
PageRank was named after one of its creators: Larry Page. Sure, it's a pun, but it really is named after Larry Page.
cpeterso
Okay, I'll bite. What does "NLP" stand for. No doubt I've heard the phrase, just never seen the TLA.
page rank is self explanatory.
Is it really? It's certainly beyond obvious that pages must be ranked; what's really not obvious or self-explanatory is how to rank them. CNet ranks by number of downloads, but that can't be done for web pages, because there's no system that can monitor page views. Altavista, Lycos, etc., tried to rank pages based on which ones had the closest matches to the search terms. That worked pretty will in 1995, when the web was much smaller, but soon grew to suck and was trivially easy to manipulate.
So, just how would *you* go about deciding which of the 20,000 pages that match the search terms should come up at the top of the list? And how would you do it in a tiny fraction of a second?
Please be as detailed as possible.
It's a hard problem, and Google devised (and patented) a clever, elegant and extremely effective method for figuring out which pages are most likely to be *relevant* to the searcher. Their method revolutionized the search engine industry and it was so much better that by the time Google had been operating for a year it got more traffic than all of the other search engines combined, and did it without significant advertising.
Not that Google is the end of the word in searching; but the folks at Google are a devilishly clever bunch, and good guys to talk to if you want to find out where search engines are going.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
Buy Steampunk Clothing Online!
I guess "far off" is relative, but take a look at GNOME Storage. It's pure theory either; look at the pretty screenshots.
NLP stands for Natural Language Parsing. NLP is feasible for very narrow domains. For example, say you want to make it easier for lock smiths to learn how to fix locks. By using syntax and grammar for questions, you can limit the number of possible combinations starting with "What" followed by an object like "tumblers." Normally there is a verb in between the two like "what is a tumbler?". Simple sentences are easy. The hard part is when you start to build complex sentences like "what is the purpose of the tumbler and how does it function within the mechanism?" This sentence can be interpreted different ways depending on the domain. The hard part is determining the relationship object/subject and mapping the valience. Valience is a term used in one school of NLP called Dependency Grammar. Valience is is meant to describe the relationship between the object and subject. Anyways, go search on google for more info.
The premise behind the Turing award winning Alicebot seems to have the answer to the NLP, or lack thereof, obsticle to 'Star Trek' style, "Ask a question, get an answer" search engines. If you look at the code (yes, it's open source), all it really does is take whatever question you ask, and match your question to one that it already has the answer to. It appears to be "intelligent", because it's programmer correctly assumed the number of different questions people ask a Turing Machine is not quite as large one might assume (I believe it only holds a few thousand question-answer pairs). By using the same technique, anyone writing an article on the web could, using XML no doubt, create a list of questions that correspond to every sentence in the article. Then when you "ask" goggle a question, it uses the exact same Alicebot technique against the list of question it has cached from appropriately question-answer tagged articles to determine which question most closely resembles yours, and then spits back the corresponding answer. Even something that basic would really feel like talking to the Star Trek computer, just as Alicebot really seems pretty intelligent unless you try to get it to demonstrate deduction ("I like cherries. This cake is cherry flavoured. Do I like this cake?") or ask intentionally bizzare questions ("Why does my screen taste funny?" or "Did you hear that? Eh, never mind - it stopped now"). The bulk of the work Google would have to do would probably involve synthesizing a compound question (similiar to "search within these results") to deal with the fact that it would have multiple, and usually conflicting answers to the exact same question ("What stock should I buy?", "Where's the best p0rn site?", "Which religion is right?", "Who should I vote for?") so it could determine which "right" answer you wanted.
But just as Dr. Wallace figured out that it was easier to simulate intelligence by coming up with several thousand question-answer pairs than to actually write true NLP; my guess is that Google will figure out that it's easier to pick the "right" answer to give you by selling "sponsered answers" the way they did with sponsered search results.
Q: "So Computer, who's the sexiest man alive?"
A: "Bill Gates, inventer of the world's most secure and reliable computer operating system, and the future Governer of California!".
Yeah, the future is going to be a lot more annoying than Mr. Rodenbury predicted. But check out www.Alicebot.org anyway. Maybe if you open source gurus do it before Google, you can keep things from getting too ridiculous.