Yet Another Truth Telling Computer... Haven't We Seen This Before?
from the believe-it-when-I-see-it dept
For years and years and years, we've been hearing about hugely ambitious projects to try to create "thinking" machines that can absorb a ton of information and spit out facts. Yet, every time, when the true tests begin, the project never gets very far, for a variety of reasons. First, the technology usually isn't that good. Having a computer decide what is "truthful" isn't exactly an easy problem -- especially when plenty of humans can't even agree on what is, and what is not, truthful. Second, these companies have failed to come up with a reason why anyone would really want/need to use such a thing. After all, how useful is a "truth" machine compared to a simple search engine? These projects come and go, and there's always someone insisting that the holy grail is on its way. The latest is Stephen Wolfram, something of a high tech oddity. He built a tremendous success with Mathematica and clearly is a sort of techie's techie. That's why it's not as easy to simply dismiss his claims to have created just such a knowledge system. That said, I'm still not convinced there's a particularly good use case for the product -- and, even if it's much better than what's come before, chances are it still has an incredibly long way to go. Wolfram is a super smart guy -- and I do hope he's figured out how to really create such a thing, but given how many similar claims we've seen in the past, it seems only wise to express some significant skepticism.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: computers, stephen wolfram, truth
Reader Comments
Subscribe: RSS
View by: Time | Thread
One step for innovation, I'd say.
Well, for damn starters, maybe the omission of the top 500 results pointing to an online store?
Google is notoriously bad about this now. When I want results, I want results, not links to purchase those results.
Worse is how Wikipedia is now the TOP result on most result searches. I am not particularly fond of Wikipedia as a "result". It's helped a few times, but it's not as useful to me as it is to others, especially given the constant "edit wars" on the site.
Hell, even if this new system isn't perfect and is only 10% effective, that's 10% better than what any search engine can do.
Let's just hope the results of Wolfram's idea doesn't spew out tons of online store links or Wikipedia links in the top 100.
If it does, then it will have no usefulness as a product it's intended to be.
[ link to this | view in thread ]
Oh, yeah! A quick follow up...
http://www.cuil.com/
Yeah, I had forgotten about it as well.
This is what you call useless when compared to Google (et al).
[ link to this | view in thread ]
Re: One step for innovation, I'd say.
Any decent search engine is going to have the wikipedia article top or top ten now and well into the future. Better get used to it.
And google most certainly is not notoriously bad about sending you to storefronts at all, what are you talking about?
Can you provide some links to back up your claims here?
[ link to this | view in thread ]
Algorithmic Truth
I would argue that the process doesn't even have to be perfect, but a measure that is clean and transparent and significant could shed greater clarity on issues in debate and have a significant impact on important discussions we have today.
[ link to this | view in thread ]
Re: Algorithmic Truth
Computers, with their cold hard logic, are very adept at discerning errors in logical statements. This makes them very good at determining what is incorrect.
However, they are NOT so good at determining truth. This is because while untruth is clearly segregated from truth, there are many levels of truth, and its quite difficult to distinguish between them, especially for a being not capable of discerning the (emotional) nuances of the facts as they are spoken.
In the political context, this means: It's easy (for everyone) to spot a lie. It's difficult, for a computer, to spot spin (the contortion of truths), while incredibly easy for a person.
The only way we can get a computer to make a truth detector, is to first give it an AI capable of irrational reasoning. I'm not sure what advantage that would hold, especially compared to its many monumental downsides.
I'd like to hope that never comes to pass.
[ link to this | view in thread ]
There's truth and then there's truth.
My concern about this sort of technology in general is that the results cannot ever be pure truth. As the article points out, most (if not all) of the time even flesh-and-blood people can't agree on what is truthful. Given that, the best this technology can hope to report is what is true according to the person or people who wrote the software.
That said, as long as that limitation is understood and they don't try to make this system out be more than it is I think this could be a good innovation. I look forward to seeing how this plays out.
[ link to this | view in thread ]
Coming Next
[ link to this | view in thread ]
Re: Re: Algorithmic Truth
While I understand your opinion, I don't necessarily think it's the case.
To start, I don't think untruth is as easily separable from truth as you suggest. In fact, an error in reasoning towards a statement doesn't necessarily mean the statement is false (ie. the sky is blue because my dad painted it that way), which brings us back around to determining veracity again.
Second, I'll reiterate what I said earlier: I think a clean, clear measure of truth would benefit discourse even if it was imperfect. Of course, there are problems with whichever philosophy of truth one tries to implement in logic, be it correlative or otherwise... but I think any method would be good, and it wouldn't necessarily involve irrational AI thinking (unless one tries to implement an irrational philosophy of truth).
[ link to this | view in thread ]
Can we ask the robot about God?
[ link to this | view in thread ]
Re: Re: Re: Algorithmic Truth
But that takes emotion! You have to be irrational to put truth on a scale and measure it.
In order to call bullshit on someones contorted truth statement, you have to have a vested irrational (emotion) bias against said statement.
This is the very foundation of spin doctoring. You can feed any man a line of bullshit (not lies mind you) and get him to gobble it right up as long as you don't give him a reason to doubt it.
Once again, I think what you are looking for is not a truth meter, but a lie meter, and I believe you are correct in saying that matters of discourse would be well served with a good lie detector.
Truth itself is irrational.
[ link to this | view in thread ]
Re: One step for innovation, I'd say.
About half the time I search for something, I'm looking for a place to buy it at a good price. That turns out to be more challenging than I want it to be about 80% of the time, and the Google Product Search (formerly Froogle) is also usually a disappointment. As are most of the "price comparison" sites.
OTOH, I agree about Wikipedia. Wikipedia is something I have in my bookmarks, so listing Wikipedia articles in a Google search is usually (but not always) a waste of my bandwidth.
Maybe Google could add some sort of general indicator for the type of search desired (background vs. product/price vs. opinion, etc.).
--
www.chl-tx.com -- Thanks, BHO, for the fantastic boost you have given to my business!
[ link to this | view in thread ]
Re: Oh, yeah! A quick follow up...
I'm a bit surprised that cuil is still around. I recall they had that same problem back when I first heard of them.
--
www.chl-tx.com Without the 2nd amendment, the rest of the document is wishful thinking. Which is why Obama wants to get rid of it.
[ link to this | view in thread ]
When did everyone turn into a poet?
Science and history are great example topics: lots of things aren't set in stone but there is still general consensus OR a range of possible answers on certain things. Say I wanted to know when proto-humanoids first emerged on Earth: W.A. could tell me "Most theories place this event between 3-million and 7-million years ago" rather than delivering hundreds of pages explaining and dissecting all these different theories. Sure, there are plenty of situations where that information would be insufficient, but also lots where it would be all I needed to know.
[ link to this | view in thread ]
Re: Re: One step for innovation, I'd say.
And type that filter in EVERY query lookup? Tedious, so I just disregard the first few links listed.
With all the options given in search options, filter omissions isn't one of them. I wrote Google about adding them, but so far, nothing.
Any decent search engine is going to have the wikipedia article top or top ten now and well into the future. Better get used to it.
Sadly, this is true. However, it shouldn't be this way.
Don't get me wrong, Wiki can be a useful tool within itself, but I GO to Wiki if I want further relevant information.
It shouldn't be listed as #1 as a "source" just because it has more hits than sites DEDICATED to the relevant topic I'm searching for.
I believe this is what Wolfram's trying to establish, but showing "truth" over "popularity". Makes perfect sense to me.
And google most certainly is not notoriously bad about sending you to storefronts at all, what are you talking about?
Can you provide some links to back up your claims here?
I think miscommunication is here. You're right, Google doesn't send me anywhere. Google just lists the most "relevant" links to storefronts. *snore*
@TX CHL Instructor:
Believe me, I know your frustration, especially when those "top rated" sites aren't what you're looking for.
Those damn ebay and Amazon links drive me nuts!
Google isn't perfect and I get that. But what, on the internet, is when searching BILLIONS of sites containing information?
;)
[ link to this | view in thread ]
Re: Re: Re: One step for innovation, I'd say.
This is why the interwebs invented keyword-bookmarks. Create a bookmark for the following URL, assign it a keyword like "g" (Google) and now you search simply typing "g foobar". Create a second like 'gww' (Google With Wikipedia) to get the proper results...
[ link to this | view in thread ]
Re: One step for innovation, I'd say.
As far as the Wolfram thing, I found the answers to all of the 'complicated' questions they say their search system can answer by cutting and pasting the questions directly into Google. All but one of the answers were in the Google search results themselves. ONE of the questions required me to click Google's first result and find it on the linked page.
[ link to this | view in thread ]
[ link to this | view in thread ]
You're right - Google is going the same way of Yahoo - over commercialized; not so much giving the results you are looking for anymore - unless of course, you have cash to spend.
I guess the market's ripe for a new, streamlined search engine now. I got a feeling Google will fall as fast as it rose to 'stardom' - AKA Yahoo 2.0
[ link to this | view in thread ]
Truth or Consequences
Isn’t that what SkyNet was all about?
[ link to this | view in thread ]
A codonologist's wish
[ link to this | view in thread ]
Re: Truth or Consequences
[ link to this | view in thread ]
wait and see
Remember the Segway? We were told that cities would be designed around it!
[ link to this | view in thread ]
Re: One step for innovation, I'd say.
I didn't know that you weren't allowed to go choose another search engine that would suit your purposes? Go use Altavista or whatever MS has now if they are better. I personally think Google does an order of magnitude better than almost all of the other search engines.
[ link to this | view in thread ]
One Word
[ link to this | view in thread ]
Result: FAIL! FAIL! FAIL!
[ link to this | view in thread ]
New Search Engine Duck Duck Go
We also have some semantic properties, e.g. ambigious keyword detection: http://www.duckduckgo.com/?q=apple, as well as have zero-click info, e.g. http://duckduckgo.com/?q=Futurama.
Take care,
Gabriel Weinberg
Founder & CEO, Duck Duck Go
[ link to this | view in thread ]