Slashdot Mirror
To Search Smarter, Find a Person?
Svonkie writes "Brendan Koerner reports in Wired Magazine that a growing number of ventures are using people, rather than algorithms, to filter the Internet's wealth of information. These ventures have a common goal: to enhance the Web with the kind of critical thinking that's alien to software but that comes naturally to humans. 'The vogue for human curation reflects the growing frustration Net users have with the limits of algorithms. Unhelpful detritus often clutters search results, thanks to online publishers who have learned how to game the system.'"
...for food?
I saw it on Slashdot, it must be true!
While I can imagine a growing job market based on this, futurists like Kurzweil in The Age of Spiritual Machines see AI coming very soon, and semantic web buffs can point to victories of semantic metadata tagging in at least some limited areas of the web. Won't many of this newly hired assistants be superseded soon?
People are better at sorting stuff before them. Algorithms, written by people, have a harder time doing what we do intuitively but can sort through more stuff. Algorithms do indeed reflect the wisdom of people, so this is a false dichotomy.
Unless we are talking about Skynet.
Except for ending slavery, the Nazis, communism, & securing American independence, war has never solved anything.
It's more a generation gap. While people in my generation are well versed on how to navigate Google and all it's side dark alleys for the gold nugget the boss is really looking for the older boss just wants it to work and is more prone to hit the "I'm feelin' lucky" button and trust what that tells them. That's where the tech snoops like us come in handy to find the obscure and convoluted information on the net. On more that one occasion the uppers have come to me to find something online because I can find it faster and more accurately.
Ask not what you can do for your country. Ask what your country did to you
...critical thinking that's alien to software but that comes naturally to humans... That seems a little out of touch with reality, there.Exam 4/C again. Maybe I'll do better this time.
Surely you jest...
I am Jack's complete lack of surprise.
Tag article "activelyavoid" and move along.
Interestingly enough, this whole thing sounds like an idea Rob Malda thought up about 10 years ago, except Brijit lacks a discussion and moderation system where experts and opinionated thinkers can vie to share their collective wisdom to enhance the content of the original article.
Support the 30 Hour Work Week!!!
Or, "1995 called, it wants its Yahoo! back."
In the absence of the mythical, impossible strong-AI, there will always be an important role for experts -- you know, thinking meat, sitting there pushing charges through neurons, having opinions about stuff -- and those experts will probably use a lot of mechanized search tools to improve the breadth of their knowledge, their awareness of knowledge, and the accessibility of information. Technology and people work together!
But you're an idiot if you take out the wetware-based BS filter.
It's coordinating all that expert opinion, and filtering out the drivel, that poses the great organizational challenge of our collective information future. Wiki-based approaches are a good first step; maybe a "trusted-wiki" like Citizendium will be the next step; it's definitely going to keep evolving. But it's long been recognized by the reasonable that if you want an informed opinion, rather than a pattern match, you ask the librarian. We've known that since Alexandria -- nay, Ur -- and it's a shame we keep forgetting.
Freedom isn't free; its price is the well-being of others.
Stumble Upon has been there for a while...
!sig
"Insane Google-fu" can be put on my resume under "skills".
What strikes me most interesting is that Google uses a set of rules in determining what it displays when queried. These rules can be changed depending on where in the world you are located, altering the results of your query. When information is passed through a set of human hands how will human bias filter into the equation? How do humans determine what is useful and what is useless information? In the end this will not be a substitute for google, merely an additional reference.
"The single biggest problem in communication is the illusion that it has taken place." George Bernard Shaw
Back in the early days of the web, it was often easier to use a web index rather than a search engine. You would go to a site like Yahoo and lookup what you wanted in a hierarchy of categories. That was often the best way to do it before search engines became more sophisticated.
There's a lot of interesting things you can do in research when you get people involved. The simplest is just hiring someone to find the information you need. I believe that a *lot* of companies could significantly increase the productivity of their developers, engineers, etc, if they maintained a pool of trained searchers that could be called upon for difficult queries (paid at maybe a fourth the rate of salaried employees). I know that I've had searches for work that took most of a day just to find that one formula I needed from 30 years worth of journal papers.
A somewhat more interesting thing, in my opinion, is all the "wisdom of crowds" stuff we see so much hype about. It's interesting because it works very well in certain cases - basically the case where the popular thing is the right thing. The main problem with this is that any search engine that shows you 10 results and then counts which ones you click, well, it's not getting your input on result #11, or 23, etc. So before anyone votes, items that happen to be near the top almost certainly stay at the top. Many good items that the algorithm ranked medium might never get voted on!
One way around this is to randomly select some less good results, so that viewers get a chance to vote for the underdogs and bring them to the top of the pile. But this pollutes results for each user, essentially making them pay a "moderation tax" by requiring them to see things that the algorithm has no reason to believe are better results.
All-in-all, social information finding features seem to be much better suited for finding things you didn't even *know* you wanted - StumbleUpon being a great example of a tool for doing that. I would imagine that this could be very useful even in the corporate sector, as many business strategies and engineering techniques have variants or cousins that are similar in function, but may be more obscure. Having the ability to see that "people who searched for X ended up wanting to know about Y too" might save me a lot of time...
AI and metadata are indeed just around the corner. The trouble is, as the article points out, that web publishers find ways to game the system. Some websites pop up at the top of the search burying the ones you actually want.
If I can guarantee anything I can guarantee that someone or some artificial intelligence will find a way to game any new system, no matter how sophisticated it is.
ref. Spy vs. Spy http://en.wikipedia.org/wiki/Spy_vs._Spy
As far as human editors go, Wikipedia seems to strike the right balance. I often include 'wiki' in my search string.
Just fscking wow.
Kudos to the guy who started the service, but the "insight" that a human can find and summarize relevant information better than a computer is hardly a surprise.
I mean, librarians and executive assistants and the like have been doing this kind of stuff for a very long time. Retrieval relevancy is a huge problem -- especially when you're talking about something as humungous as the internet. There's so much crap out there, it's likely always going require a human to do good executive summaries.
Cheers
Lost at C:>. Found at C.
Expect to lose your job soon after the paperless office arrives. It's always just around the corner but something human gets in the way every time. AI will be much the same.
http://slashdot.org/comments.pl?sid=216934&cid=17629948
Human-based KNOWLEDGE searches vs. automated INFORMATION searches.
Very fun, charming little movie, all about the perils of automation. Check it out, even if you have to use up your next Netflix delivery. Worth, if for nothing else than seeing Spencer Tracy and Katherine Hepburn onscreen again. http://www.imdb.com/title/tt0050307/
What's worse: human bias towards a particular resource (i.e. like cooking site X, but not site Y) or limitations in contextual based results from computer algorithms?
ilovegeorgebush
Yahoo: humans organize content.
Google: magical search algorithm organizes content, gets it right sometimes.
We're back to the Yahoo! model because people have figured out how to game the system, namely Google, without adding content that's important to the searcher.
I welcome this. Our computers can't yet come close to matching our brainpower.
technical writing / development
But what if the system being "gamed" is a human-based search engine? Since the publisher must fool humans anyway, the "unhelpful detritus" in the end users' results will blend in. Even if there are fewer false positives, those that remain will be harder to eliminate.
And the new AIs will be powered by fusion, and will drive my flying car for me.
There is a skill in quick filtering and searching for information online. Ask two people to find a fix/workaround to a software error in an Off the shelf product and they will take various paths to their respective solutions, if not the same solution. If the initial search doesn't turn up enough hits, you can immediately reason what is an alternative search string to use, replace "Error" with "Crash" or "Bug" or "bombs" or "Issue". You can refine based on the results on this pass.
Can that decision tree be coded? Sure can, but the physical labor cost to performance ratio is probably far better than the research/development cost to performance ratio in the near and middle far term. And with people you can greatly different methods as opposed to a designed "AI", being designed, it already has constraints in it's methods based on the design.
Some day....but I don't think "AI" is even remotely close to that level of sophistication.
Sehr geehrter Toilettenbenutzer!
Back in the heyday of free hosting services like Geocities and Fortunecity, small sites (mainly by and for fans) didn't rely a whole lot on search engines to drive traffic. Much more common and trusted were instead Webrings. For those who never partook: a webring is a loose community of related websites. It was moderated by a handful of people, and each site would put a little Webring script at the bottom of their page(s). This allowed users to surf between related content without having to go to some external website. It built more trust between the websites.
While I have not RTFA (this is Slashdot, after all), the summary makes it sound like the combination of Webrings and "Top X" lists, both of which are used much less now and don't carry as much weight but still require user interaction on a grand scale.
I'd be interested to see how this kind of search engine turns out- however, you also have the problem of "majority think", so searching for, say, evolution might have a first result for a page "debunking" it. But then I browse at +4, so I shouldn't complain.
There are places for that, like /. and everything else, but over time, it can still evolve and improve, I guess. However, I feel that the real problem is more of how shallow search algorithms are designed in the first place by any site that uses one in some form and not just search engines. (For me, at least:) It is rarely the case that I want my query searched in the same, consistent manner. The common search engine's advanced search is sort of on the right track, but the information of the internet would really feel a lot more at my fingertips if it, and search algorithms everywhere else, had a lot, lot, lot more depth. I would guess that search algorithms really can be improved and that this would be a much smaller issue of the internet than it is today.
Using humans to rank or select is not exactly old.
:-P
What distanced Google over purely statistical keyword search engines like Altavista was actually the use the human ranking implicit in the human-created URL links between pages[1]: the application of citation analysis[2] to the web.
Oooh, look, it's been invented again! Hooray!
[1] http://en.wikipedia.org/wiki/PageRank
[2] http://en.wikipedia.org/wiki/Citation_analysis
Wikia shows the problem with this approach. Coverage of Star [Wars|Trek|Gate|Craft] is extensive. Coverage of, say, bank regulation is nonexistent. If you want to find out how we got into the subprime mortgage mess or what to do about it, Wikia search is totally useless. That's what you get from volunteer editors. Wikipedia does better, but most of the good contributions were made years ago.
Today, you pay the editors, or you get fancruft.
It's amusing that the author of the article feels overwhelmed by The Economist. That's a very well written magazine with good reporters; they had the only reporter in Lhasa when the Chinese clamped down, and they have a good analysis this week of the issues surrounding derivatives. If this guy can't handle The Economist, his organization's answers will probably be dumbed down to the level of, say, "People". That level of crap one can get for free, from many existing sources.
Remember Google Answers? Nobody really cared, and Google shut it down.
There's a whole industry of expensive, small-circulation specialist newsletters, but those are niche operations run by specialists in narrow fields.
The Goog algorithm works as long as you do the same search that everyone else is doing. Since large most people have the same interest at any moment, it works. God forbid people start thinking independantly.
I run a virtual reference service for a provincial public library collaborative . Our stats are off the chart. We are running at triple the customer traffic that we had expected. So, essentially, lots of people already know that a person can find them stuff way better than an algorithm. There are two basic problems with this:
1. is that most people hate looking for things
2. is that most people are lousy at looking for stuff on the web.
Our solution, of course, is that we have institutions full of people what *do* like looking for things and *are* very good at it. THose people are called librarians.
Ciao, Dcobbler.
Yea, but some people just want to ask the librarian where to find the book they're looking for.
Prediction: The real iPhone killer is going to be sex robots from Japan. Think about it.
Sounds a lot like Dmoz to me! http://www.dmoz.org/ The problem with such sites - the Internet is just too BIG!
In fact, it's a basic theorem that given sufficient time, human-level intelligence can always beat any system with less than human-level intelligence (aside from trivial cases like a complete firewall). This is because the human's theory of mind can fully encompass the lesser system (so you can understand how it works), while the reverse is not true. Computers can only beat humans at chess when the match is played with a time control.
This doesn't mean that a computer system can never be good enough to solve this problem. However, it does mean that if you could build a computer system that could solve it, then it would insist on being paid.
It also doesn't mean that using human-level intelligence will always solve this problem. Humans can still be beaten, they just start on a level playing field. Hence it's pretty much inevitable that some people will still find ways to game the system.
In the ancient days, when anyone wanted to know where the best restaurant in town was, they'd ask the guy on the street. Along came the great uniter of minds, and people could find what they wanted by asking the whole world for it. But now, we think asking the whole world is not good enough... so we go back to the guy on the street (who's probably renamed himself to "StreetSearcher" now).
We're back to the Yahoo! model because people have figured out how to game the system, namely Google, without adding content that's important to the searcher.
It's not hard to throw out most of the bottom-feeders. We do it. The crowd at Search Engine Watch (which, despite the name, is all about advertising, not search quality) is writing me angry messages for doing that. Now that we've demonstrated that 36% of Google AdSense advertisers are bottom-feeders, they know they're being watched. Some feel they're being targeted.
Bear in mind that most search requests are really, really dumb. That's what Google has to answer. In fact, most Google search requests don't hit the search engine at all; there's a cache of common queries and answers in all the front end machines, and a sizable fraction of requests are answered from cache.
For about 5-10 years until the search algos get smart enough that they can beat a human.
Seriously how far off is that? We're not talking about the singularity though it would have to be pretty close to the turing test. Ask the search engine a question, not quite what you want? Ask to narrow it down, if you cleverly crossreference the results with the narrowing question it should be fairly feasible to do comparably as well as talking to a live person who knows how to search. I'm not sure I'd want to build, or invest in, a company built on the premise that they'll make up their investment before computers become better searchers than humans in a few years.
I stole this Sig
The summary mistakenly wrote "Unhelpful detritus" when it should have had "sea anemones of the web who need a piece of reinforcing rod steel swung at their shins."
Hope this helps.
Skim the summaries, and occasionally there's something worth clicking through. It's not a new idea at all.
It also seems like a bad idea, if it's based on this premise:
And it's much easier to game human systems than algorithmic ones, I expect.
Don't thank God, thank a doctor!
Unfortunately you're betraying your own point by making a comment involving "critical thinking."
"Where is the wisdom we have lost in knowledge?
Where is the knowledge we have lost in information?"
--T.S. Eliot
Unfortunately, I have no information to provide at this time.
The publishing world has been doing this for decades now...hiring people to track down obscure references, sources, photos, drawings, paintings, any kind of data.
I doubt most of these new "online" ventures will actually go out and hire qualified "searchers." My guess is that they will rather just pick up a bunch of freelancers looking for part time jobs and who probably know even less about the subject of what you are trying to search on than you do.
On a side note, what happens when the SEOers figure out a way to pay off the online search guides???
The critical human thought phrase has been struck down, though I think for many of the wrong reasons. A long time ago (car analogy incoming) people used to work on their own vehicles much more so than today. The onboard computer stopped a lot of that, and general complexity stopped more. With home computers and the Internet both problems exist and for many people (until this recession hits hard) it is cheaper to pay someone else to find stuff than to figure out how to find it themselves.
It's not really difficult, many of those sufferers know how to use a library, which is the real world equivalent of searching on the Internet. (not that the Internet is not real world) Most people were taught how to use a library in their school days and that usage has not changed much with time. The usage of Internet searching does change, and there are multiple ways of doing it. People who are not interested in learning new ways will always just say it is too difficult.
Using boolean modifiers or advanced search is always there, people just don't use it. They also don't fix their own lawnmowers or other things. They just replace them or pay someone else to do the 'hard' stuff. There is enough information on the Internet to allow anyone to learn to protect their home computer from infections and malware, yet it still is a problem.
The human problem of search engines will NOT go away, it can only be made to look less with smarter UIs. A tag cloud system of bookmarking could be used to refine search results but would not work in all cases. The URL history with timestamps might help, but not in all cases. Analysis of search results and those pages actually visited might help narrow the criteria to personal bias but not in all cases. That is why the operator has to be smart enough to know what they want and don't. The Internet does not come with your very own personal cruise director to make sure all goes well. People just believe that it is supposed to be easy because they want to do the cool things that they hear about on television and from their friends etc.
Perhaps one day the interface will be fast enough to be considered good when our brains can be plugged into the computer itself, something like The Matrix, reducing click delays and reading to milliseconds. Until then, teaching people how to use complex search strings will help reduce the angst and pain.
"cars +toyota -hummer 2005" aobut 2.98M hits
is better than
"cars 2005" about 19 million hits
but you have to teach people that those extra characters really REALLY do help.
If people don't know how to use a soldering gun, please don't give them one... or something like that. Oh yeah, car analogy: you apparently can't drive on the streets of the USA legally without a license, which you cannot obtain without demonstrating proficient control of the vehicle.
Support NYCountryLawyer RIAA vs People
It's one thing to say humans are good at looking at what they are looking for themselves. It's another thing to try to build an industrial activity around the notion. Humans are NOT reliable and are prone to zoning out when they are performing repetitive tasks. This would indeed be a repetitive task. And in spite of all the warning dialogs people see, eventually, people stop reading them and click "yes" to everything just to get it out of their face. Given that this aspect of psychology is pretty well known and accepted, what makes them think it doesn't apply to filtering web content?
Every time I search for a description of some IC chip, especially an old, rare or highly specialized one, there are THOUSANDS of fake chinese cover-up distribution companies' websites with autogenerated part indexes. No, those aren't parts they actually have. Those are all pages generated using a list of bazillion different chip names, some of them nonexistent and generated just in case such a version could exist in the future, to catch any attempt to search for actual information. No matter what you click, you get an "availability inquiry form" with the part name pre-filled and that's all. Maybe they think this will make someone buy from them, but they make finding anything close to impossible. Some even go as far as autogenerating "search" pages with different part names "submitted" (really a crapload of invisible HTTP GET links in footers of other pages), but of course no results or links to inquiry forms. Others link datasheets for EVERYTHING - to a ".pdf" file redirecting to a "Sorry, we don't have this" page with 200 OK status, or an empty PDF. There goes your chance to find datasheets for anything less popular than a MAX232.
I'm all for shooting them, running them over with an 18-wheeler and hanging from a tree as a warning.
This is Slashdot. Common sense is futile. You will be modded down.
I have often wished that searching was more tuned to the way humans ask questions. Ask Jeeves attempted this way back in the early days of websearch, and Google is very good at it now, but not with the types of things I sometimes want to search for. Sometimes you want to find something when you don't know what it is called. Your best bet is to type in a few adjectives and hope something comes up.
For example, I was humming a tune the other day. I know it was an Aerosmith song, and I can hum a few bars, but for the life of me I can't remember any of the lyrics. I wish that I could google for "that song by Aerosmith, the one that goes da daa dada..." and Google would be able to answer it, much like a friend would. Another example is when I see a commercial on TV for a store, and sometimes I'll google a phrase like "such-and-such company's commercials are funny", hoping to find an online forum where other people who agree with me have previously discussed this funny commercial. Doesn't always work, naturally.
The beauty of the way humans interact is that we can process a lot of information very quickly without having all of the information. If I'm working on a project and need a tool, I can tell a buddy, "Hey could you pass me the whatcha-ma-callit, you know the turny-one?" and he can probably deduce that I want a 3/8" socket wrench based on context clues. I hope that one day, Google search will be like that.
Weren't a few early search engines like AskJeeves and those just a basic spider with a load of actual humans filtering popular searches?
What he can't kill, he has sex on. Trent.
You search for anything that is a product on google and you get pages upon pages of basically adverts. And adding "review" to a product hardly returns "real" reviews, just customer reviews from online stores. One must be very selective when searching for anything google makes a dime off of.
http://www.dmoz.org/ This is nothing new, the concept/technology has been around for years.
Kurzweil, unfortunately is a bit of a nutbar. There are many things that never turned out quite like how people predicted.
He assumes that AI is a known (i.e. understood) problem, AI is not understood, not by a longshot.
The only reason that chess cannot be solved completely is because there are too many solutions for today's computers to search out all the solutions, and figure out the optimal solution. Computers can only go so many moves ahead, especially when you factor in the time constraint. However if you look at a game of checkers, we have gotten computer algorithms to a point where it is impossible to beat the computer. The best you could do is tie. I imagine that the same will be true eventually for the game of chess, once computers are fast enough.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
Wow, I remember back in the day when we only had one search engine and it was human powered with real links to real content. It was called Yahoo!
Well, it's also a problem of understanding the results, not just of one of knowing how to use Google.
1. Let's say I'm interested in legal advice, for example. I know how to use Google, but (A) it will take me disproportionately more time to understand it than it would take a lawyer, and (B) I'm still not sure if I understood it right, or if the person who wrote that does. Sometimes Google isn't the Alpha and the Omega. Sometimes I'd rather pay a lawyer to search for me, than trust my l33t operator-combining skillz and Google.
Not only I can see a manager doing the same, I can respect him more if he does. If you suspect you're not qualified enough to understand the points of this newfangled Snake Oil 2.0 Enterprise Edition framework, _don't_ Google personally. Delegate to someone who understands it.
2. Basically let's put it like this: bosses had secretaries for ages, and not because some "generation gap" makes him unable to use a typewriter or lick his own stamps. It's just that (supposedly) his time is worth more. If you have a secretary spending a hour on menial tasks, it's cheaper than if the CEO paid millions per year does it. Even if he's exactly as fast at running stuff through the photocopier and putting it into envelopes, an hour of that for him costs more money than an hour of the secretary's time.
Way I see it: same here. Even having the mad skillz, and understanding the topic (e.g., because it's something trivial), searching for some topics still is a major time sink. Yes, it'll be faster for someone with the mad skillz, but it's still a time sink either way. At least theoretically it's cheaper if the secretary does the searches, than if you have the CEO spend half his day googling.
A polar bear is a cartesian bear after a coordinate transform.
Where are these semantic web victories? Slashdot's tagging system even uses human "assistants" and its still a miserable failure as far as finding information goes.
If you want to wait for AI go ahead. It's much like waiting for cheap fusion power though. I'm sure it will happen someday, but I'm not holding my breath.
The people who want to "just ask the librarian" can use online equivalents such as Uclue paid Q&A/research, Ask Metafilter, Wikipedia Reference Desk etc.
Paid search is always going to be a niche business, because most people don't want to pay, and because it doesn't scale as well as algorithmic search. But for those who want to use paid search (such as Uclue), it's a valuable service.
Paid Q&A/Research
So... to find this person, should I just use Google or do I have to find a person?
This space up for sale.
Your statements are based on the faulty premise that computers will never be able to learn, or at least learn as well as a human. I would never make such an assumption. People have an elevated sense of themselves. It's bullshit. We're just fancy bio-computers. Once we understand how that really works, we're be able to replicate it and (probably) improve on it.
Are agnostics skeptical of unicorns too?
Hey all!
:) By pairing you up, they filter out the fact that someone is polluting the searches too, so it's really smart that way.
:)
I personally think this is a great idea. Computers can only filter content to the best of an algorithm and frankly, standards are very poor on who codes what, how and how clean code is. Going human is definitely the way to go.
If anyone here has tried Google's Image Labeler you'd know what I mean. They pair you up with someone on the net, you tag images leaving out all the most popular links and when you both match one up, you get points and move on to the next one. This helps Google filter out results and add the most HUMANLY PERTINENT results.
Give it a go: http://images.google.com/imagelabeler/
I really think this is a great way of getting more pertinence out of results and will benefit the overall greatness of our search engines. That, and Google has found a way to get the public to do what it's been hiring people to do for years already. Way to get people to work for free Google!
Or, if you prefer a religious reply: we're the result of God's divine intervention, and can never hope to match His work
No, they're based on the underlying logic. Your statement, on the other hand, is based on failing to specify what "learn" means, and hence avoiding thinking about the subject, combined with not even bothering to read the post to which you are replying.
When you have two given systems A and B, then the following statement is either true or false:
System A is able to understand how system B operates
(This statement is obviously transitive. We do not know whether it is commutative, that's one of the big unsolved questions in the field)
We call this a "theory of mind": if your theory of mind is strong enough, then you can understand any systems that are weaker. A human has a theory of mind that is strong enough to understand how any Turing machine operates. We have never been able to construct a Turing machine that has a theory of mind strong enough to understand how a human thinks (partly because we don't know ourselves), and we're not sure whether that's even possible.
This part's too complex to write out here, but it is fairly straightforward to show that in order for system A to fully understand system B, it must effectively be at least a system B. In order words, if you can construct a computer that can fully understand a human, then that computer is a human, for all practical purposes. It has desires and emotions and motivations like a human, and it has (or should have) all the privileges of a human, and it is not going to want to be your slave. You're going to have to find some way to pay it if you want it to work for you. Hence, you have accomplished nothing by building it - you could have just used a regular human.
You are completely clueless as to whether or not that is really true. People who actually know what they're talking about haven't been able to figure it out in the past couple of hundred years, so you certainly haven't. It is an open question as to whether or not human minds fall within the scope of the Church-Turing thesis, and hence are equivalent to computers, and we won't know until we figure out how the brain actually operates.
Are agnostics skeptical of unicorns too?
True, yet completely missing the point. A computer today can be beaten by any human in the absence of time constraints, because the human understands how the computer operates and can devise a strategy that builds a trap which the computer cannot see. For example, if we take the primitive case of a computer that looks only five moves ahead, you simply have to build a trap where the next five moves all look very good (let the computer capture a piece on each of them, or something like that), and you checkmate on the sixth. The computer can't do the same thing to the human because it doesn't understand how the human thinks about chess.
Chess masters can do this to the current computers - it just takes several hours (or days) to figure out, which is too long for games played under normal time control.
I fully read your post and understand it. I'm sorry you have such a limited imagination for such things but I'll try and make this simple for you: Theory of mind is bullshit. Sorry but that's for idiotic academics that have such limited imaginations that they have to put an artificial restriction in place. Eventually, humans WILL understand how the human brain works. Once you know how it works, there's always the opportunity to improve on it. It's just that simple. Once again you fall into the trap that humans are somehow magical beings, that we're somehow unknowable...that we're like God. We're not. We're biological machines with various abilities and limitations just like any other animal on this planet. Once you understand that, then you will understand that we can learn what we truly are and how we truly work and make something better. It may not even be software in a circuit. OPEN YOUR FUCKING MIND.
Are agnostics skeptical of unicorns too?
But yes, given time, I have no doubt that we will create some sort of artificial life. I doubt it'll run on the X86 instruction set though
Are agnostics skeptical of unicorns too?
Those people I work with that I've thought were just lazy bastards who couldn't be bothered to lift a finger to search something and ask me instead... were actually geniuses who had the right idea.
All kidding aside.. I've found an increasing number of people who seem to think I'm their human google proxy. When someone asks me a question that I can cut and paste into google and get 10,000+ results, the first dozen plus pages all clearly describing the issue and usually fixes, I get pissed. Of course these are the same people who can't be bothered to spend any time reading the docs I had to spend time writing (or I get dinged on my review).. and when I point them to the docs vs spending my time explaining it to them instead.. I get dinged for being "difficult to approach for help," or some other BS. To these people, the word "help" also seems to actually mean "do this work for me I don't want to do."
- My favorite error message: xscreensaver, running on an old Sparc 5 w/ 8bit color: bsod: Couldn't allocate color Blue
"That's not as unorthodox an idea as it might sound. Lots of professionals have always had assistants whose specific purpose was to research, proofread, fact-check, etc. Doing internet research is now just another skill for those types of assistants."
India called. They want their career back.
The article is misleading: Google's rankings are based entirely on human evaluation. The signals Google uses -- like the number and origin of incoming links to a page, or the specific title and headings the author uses to summarize their content -- are entirely the work of humans trying to make their content useful to readers. Google's algorithms in effect just aggregate this human-created information into a convenient form. (True "algorithmic" search that doesn't use human-generated signals is the problem of general AI, and as-yet unsolved.)
Google also demonstrates that since this information is from a large number of distributed individuals, it's fairly unbiased, reliable, and difficult to game overall. Google can however be gamed (biased) in a narrow specific area through a targeted attack, which I suppose is one issue these competitors are trying to solve. Google's goal of course is to increase the sophistication of the methods it uses in order to filter out these attacks. There is a significant learning curve here, which ultimately is one of Google's advantages over competitors. (Ironically these spam attacks are therefore good for Google, because they keep the search problem hard and allow them to differentiate themselves against competitors.)
It should be recognized that any competitors with hand-edited results pages (Mahalo, etc.) are subject to bias of a different sort: Human bias. If a single editor is creating results pages on specific topics ("abortion" or "sex" or "Taiwan independence" or "George Bush" or "Adolph Hitler" or "Mormonism", etc.), one is naturally concerned about the inherent bias of the editor/author. Wikipedia's approach to this problem is to allow a broad set of people to edit the pages, in the hopes that biases will be detected and ironed out. Since none of these "new" search engines I'm aware of allow this type of openness, it seems to me they run a real risk of just being crappier versions of Wikipedia.
"Robotics is an entirely different issue. There are already robots that can walk and run. Granted, there are none that can handle obstacles as well as a human but these problems will get worked out in time. There's no reason to think otherwise."
Robots are definitely more able than i am getting from toronto to new york. Walking is a horribly artificial challenge especially bi-pedal. Research on bi-pedal robots only started a few years ago (in earnest) and we have already seen big advances).
Now where timeline and evolution is concerned it took evolution 400 million years to get from arthropods to mammals.
Humans made the carriage in 100BC, 1800 we had steam railways, bicycle in 1860, motorcycle by 1890, airplane by 1905, rockets by 1926, 1950 we make computers (5000ops), 1960 hardrives made and we launch something into space, 1970 ram invented we hit golf balls from the moon, 1980 Dos 1st pcs pioneer 11 reaches saturn, 1990 internet invented hubble telescope launched, 2000 space station half built supercomputers reach 10,000Gflops.
Evolution has got nothing on our current and increasing speed of advancement. I'm sure we'll surpass ourselves soon enough. The human brain does 100Pflops... current super computers are at 1Pflop... meaning computers will overtake the human brain in processing power before 2020. I'm sure cognitive ability will follow shortly after.
"Computers are fast but stupid, humans are smart but slow. The only way a computer can be smart is if a human makes it smart. And even then it is only as smart as the person an make it"(slightly reworded) The thing that a computer helps us with is not the searching for complex information it is finding the simple thing that we are looking for and eliminating the things that we dont want to waste time with. A computer is nothing without a person and a person is slow without the computer.
The automated pagerank algorithm IS human-driven. It ranks most highly those pages that are linked to by other pages. But who put those links in place? Humans of course!
When I (or most other people) write web page content, I link to other pages that I've found to be the best - I don't link to trash. By looking at the pages I chose to link to, Google et'al are using my expertise in that subject area. Pagerank is merely a system of counting votes cast by experts in a given area.
The clever part is that it does this without the cost of paying the people it "consults".
The not-so-clever-part is that there are biassed experts out there - and people who stand to make money by providing bad links. Pagerank's complications are actually ways to try to ignore people who are biassed in how the "vote" for pages by linking to them.
But any system of real human consultants such as described in TFA is liable to similar biasses. If you recruit a person who claims to have been abducted by aliens into your expert group - then links to flying saucer articles will be found in unreasonable quantities and given unreasonable weight - just as if that person had covered the net with lunatic ravings that Google would have picked up on. You STILL need a way of sorting the "good experts" from the "bad experts" - and that's just as hard for both approaches.
The problem is that a computer could think 30 moves in the future, provided it was fast enough. I don't think that a person could ever think 30 moves ahead. There comes a point when the computer will just plain be better than the human. They've even got computers that beat just about everybody at poker. Probably won't be long before we have computers that will beat even the best people.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
Of course that all breaks down when there is a misunderstanding, how ever goodwill and good manners will generally resolve this. Now trying to create programs to mimic this, is a sure recipe for GIGO, garbage in, garbage out, a whole lot of frustrated people and a bunch of broken keyboards.
Chaos - everything, everywhere, everywhen
Yeah, because using people for Searching has gone so well for Yahoo hasn't it?
Add -https to your google search to get rid of the salesmen.
.com sites which are usually commercial, for-profit enterprises.
.org and .edu. These are usually not trying to sell you something....
A test I did just now with google....
'software' Results 1 - 10 of about 1,700,000,000
'software -https' Results 1 - 10 of about 176,000,000
Filter out
'software -.com' Results 1 - 10 of about 72,400,000
Even better, restrict your search to
'software +.org +.edu' Results 1 - 10 of about 70,900,000
Advertising is really unecessary since we have:
1) The Internet
2) Phonebooks
3) Your five senses
4) Word Of Mouth
Hope this helps....
Because nobody was paying Mother Nature to eveolve humans, the process was pretty haphazard and filled with a million years of errors. We might do significantly better by paying attention to the details and culling obvious errors far earlier.
It could be fun to set up some "Minsky Awards" for AI's that manage to destroy themselves, to match the Darwin Awards at http://www.darwinawards.com/.
So if I can understand a recipe for tortillas, I'm a tortilla? Hardly. The ability to understand or even master a system does not make one the same as that system: comprehension is not identity.
And frankly, there's nothing "open" about the Church-Turing state of humans. We're far too error prone to even approach a good computer's ability, we forget too easily, and our fundamental neurology is not binary. It's a dynamic 3-D analog network, and the neurons are not digital. They're impuse transceivers, not well modeled by binary state machines.
Go back to your basic biology, and examine the numerous reasons why a squid axon is not a digital system.
It is one thing to buy into pop science --- we all do stupid things at some point in our lives.
It is quite another to buy into pop science by RAY KURZWEIL --- now that's really hopeless. What has the world come to.
>>>"But isn't AI and metadata just around the corner?"
Yes. And it's been "just around the corner" since the 1970s (along with battery-powered cars and flying cars). There's a difference between predicting something will happen, and when it will happen. So far the task has proved itself far more difficult than people originally thought (which is why A.I. was predicted to happen in 1980 - and yet still had not happened).
The government is not your daddy. Its purpose is not to raid middle-class neighbors' wallets and give it to you.
Your both wrong. In order to make tortilla you need to understand the recipe yes? NO a machine can be made that can make tortilla without understanding the recipe. So we are left with Searle's Chinese room argument. I think it more interesting that I know what a tortilla might be without actually knowing what one is. I know it is some sort of mexican food? I know you can eat it? I know they give you bad wind?
As ahabswhale already pointed out, "your perception defines your reality". The first step of teaching computers to think will likely be teaching them how to acquire, analyze, understand, and react to incoming data, much like we react to input from our eyes, ears, noses, etc. Maybe once we've solved that issue, adding intelligence into the mix will be a breeze, but we're still a ways off from achieving even the first step.
Amazing - the idea of people whose skill is to find information in a world where there is so much information that it's impossible to know everything. We've been around forever - they call this job Librarian, Researcher, Information Professional, etc. It's not new, and we're damn good at what we do. We've been doing this longer than you've been on the Intertubes. Seriously, if you think Librarians do nothing but check out books, it's you who are way behind the curve.
The problem with this is that people would freak out over the privacy implications well before Google had enough contextual information to answer queries like this. :-(
ttuttle is a rankmaniac
How could you possibly call this a theorem? You're using terms like "human-level intelligence" that aren't even well-defined, and you haven't specified the parameters of the game. Are you saying humans can win any game against a computer? Not true. There's no rigor here.
We have the same problem as before. What is a lesser system? Are you referring to some kind of Kolmogorov complexity? One problem with defeating computers at certain games (like checkers, which has been completely solved computationally) is that the human brain cannot store very much abstract symbolic information. If you ask an unaided human to play checkers against a suitably programmed computer, the human will be unable to keep track of enough information to avoid mistakes. It doesn't matter what your "theory of mind" is (whatever that means) if you can't remember enough of a system to understand how to exploit it.
Not necessarily. The problem of gaming search results is more of an economic question than an AI-theoretic one. The goal is to arrange matters so that any successful strategy for gaming the system realizes limited benefits, and requires a large investment in resources. Brute force computational solutions often make economic sense, because they can scale better.
"Your notation sucks!" -- Serge Lang (1927-2005)
This seems odd. You're calling me "wrong" without contradicting the points I made. Some level of understanding and flexibility is pretty necessary to organics and artificial intelligences, or they're inflexible and relatively uninteresting.
And by the way, it's not the tortilla that gives you wind. The tortilla is just a flat soft bread, used to wrap up food or grab food with, lots of cultures make versions of them. It's the cheese and beans you put in the tortilla to make a burrito, just as a bit more thought to your argument would have made it more interesting and effective.