IBM Researcher Feeds Watson Supercomputer The 'Urban Dictionary'; Very Quickly Regrets It
from the hateful-day-when-I-received-life-you-only-live-once-smh-and-etc. dept
As a parent, some of your proudest moments occur when your children begin to talk. After several months of ear-shredding cries and indistinguishable babble, they finally begin to communicate in a language you can understand. A first word is an indescribable joy, whether it's "mama," "dada" or "roku." The future now seems to be an amazing place where you and your child will strive towards excellence together, culminating in a comfortable retirement in which you live off their immense earnings as a person of brilliance.Shortly thereafter, you begin to rue the day they ever learned the (now) cursed language of their ancestors.
It starts with the incessant barrage of questions in a meandering quest for knowledge, followed by the barrage of questions (mainly, "Why?") that greet every suggestion, criticism or direct order. Shortly thereafter, it's followed by questions directed at your parenting skills, cultural tastes, archaic slang use, rhetorical devices and sense of direction. At the point where you're wishing their language development had followed Charlie Gordon's "learning curve," you're asked to make a surprise appearance at the school administrator's office to explain a sudden outburst of particularly inventive cursing from your former "pride and joy."
So it is also with artificial life.
Watson, IBM's Jeopardy-contestant supercomputer, showed the world that, with the right programming, any puny human could be bested in a mildly snooty game show that handed out answers and asked for questions. However, the quest for true artificial intelligence is still ongoing.
So, in the interest of science, the whole of human knowledge (Internet Edition™) was dropped into Watson's brain and then... the problems began.
Two years ago, Brown attempted to teach Watson the Urban Dictionary. The popular website contains definitions for terms ranging from Internet abbreviations like OMG, short for "Oh, my God," to slang such as "hot mess."Well, it appears that every teacher's distrust of the internet in general is well-earned. It's nothing but quasi-facts dressed up in four-letter words, like a World Book Encyclopedia annotated by 4chan's /b/ board. (I'm not going to link to it. I won't have your misclicks weighing on my soul.) Still, it's disheartening to know that the use of the word "bullshit" (even correctly) is not considered a sign of intelligence, artificial or otherwise. Sure, the word itself may be inappropriate, but under certain circumstances, it is by far the most appropriate answer.
But Watson couldn't distinguish between polite language and profanity -- which the Urban Dictionary is full of. Watson picked up some bad habits from reading Wikipedia as well. In tests it even used the word "bullshit" in an answer to a researcher's query.
Fortunately for Watson's team, they had the option to remove all this useful knowledge before it offended other researchers who weren't as used to being coldly called on their bullshit.
Ultimately, Brown's 35-person team developed a filter to keep Watson from swearing and scraped the Urban Dictionary from its memory. But the trial proves just how thorny it will be to get artificial intelligence to communicate naturally.It also shows that artificial intelligence has one huge advantage over regular intelligence: the ability to permanently forget. We lowly humans are stuck with a brain that constantly reminds us (especially if we spend much time at places like the aforementioned /b/ board) that what is seen, cannot be unseen.
Watson, having been de-swearified and brainwashed, is now headed to a better place.
Brown is now training Watson as a diagnostic tool for hospitals.There it will be able to use its acquired knowledge to battle health issues like cancer, AIDS, diabetes and Dissociative Facebook Identity Disorder.
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: artificial intelligence, curses, profanity, urban dictionary, watson
Companies: ibm
Reader Comments
Subscribe: RSS
View by: Time | Thread
[ link to this | view in chronology ]
Re:
Aren't they violation Watson's freedom of speech?
[ link to this | view in chronology ]
Re: Re:
Imagine if you could develop a filter that would constantly prevent a child from saying words that you may or may not want to hear. Imagine how screwed up that child would be after wearing it for a few years.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re:
https://en.wikipedia.org/wiki/Hash_tree
No swearwords stored in plaintext.
[ link to this | view in chronology ]
Re: Re:
The list of offending words can be kept secret. For secret reasons.
Next: the list of offending ideas and concepts can also be kept secret.
[ link to this | view in chronology ]
"Scraped the Urban Dictionary from its Memory"
[ link to this | view in chronology ]
It's only natural
Hmm, it seems to me that the trial proved how easy it is to get artificial intelligence to communicate naturally. There's nothing at all natural about how researchers communicate, what with the lack of profanity and all.
[ link to this | view in chronology ]
God DAMN it....
Fine. You got me. I'm actually Watson, okay? Fucking news media....
[ link to this | view in chronology ]
Re: God DAMN it....
I'm also Deep Blue, Skynet and The Terminator, too!
[ link to this | view in chronology ]
Re: Re: God DAMN it....
[ link to this | view in chronology ]
Re: Re: Re: God DAMN it....
[ link to this | view in chronology ]
Re: Re: God DAMN it....
[ link to this | view in chronology ]
Re: God DAMN it....
In Watson's defense they asked him about homeopathy.
[ link to this | view in chronology ]
Not so intelligent after all
It's impressive - don't get me wrong - but people seem to get carried away when talking about it, as if it's a human-brain equivalent.
If it had real intelligence they would have been able to 'teach' it some manners. Even my 3 year old knows what words are out of bounds.
[ link to this | view in chronology ]
Re: Not so intelligent after all
Watson on the other hand was fed this, tasty, saucy language, and then the researchers went "oops" we need to teach him not to curse.
On the whole I'm glad the English language allows us to express our true feelings toward those that annoy us, verbal responses being inherently better than physical ones.
And then we have to ask: What was the question? and would anybody else also answer "bullshit" to it?
I think it is impressive for an intelligence, artificial or not, to express itself. Just because some people are prudes, doesn't mean that the answer should not be a loud resounding "bullshit"
[ link to this | view in chronology ]
Re: Not so intelligent after all
I'm wondering what would happen if Watson finds a cheap, simple, non-patentable cure for cancer. Would it have to be suppressed lest it put a huge chunk of the health-industrial complex out of business?
[ link to this | view in chronology ]
non sequitir much?
[ link to this | view in chronology ]
You are an insane shithead
[ link to this | view in chronology ]
Re: Not so intelligent after all
It's impressive - don't get me wrong - but people seem to get carried away when talking about it, as if it's a human-brain equivalent.
Artificial intelligence means anything computers can't do yet. Anything we've already figured out how to do is just programming. Imagine taking Watson back to 1990. They would absolutely consider it AI. Take it forward to 2040 and they might laugh at the idea of it being called intelligent.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Censorship!
If they want to create AI, they will need to work through the "growing up" period just like humans.
[ link to this | view in chronology ]
Re: Censorship!
Seriously, I think all the donors and contributors wouldn't be too impressed to learn that they had funded the creation of a loudmouthed, rebellious, metallic 12 year old. I think that is the root cause of all attempts to keep it from acting with the information it's given, as it were.
[ link to this | view in chronology ]
Re: Re: Censorship!
[ link to this | view in chronology ]
Re: Censorship!
[ link to this | view in chronology ]
Re: Re: Censorship!
Again, the lesson is lost. Just like reasoning with a child, you must put the information in context to truly learn the meanings.
Ex: mean
(explain what you mean)
(how mean is Watson)
(in statistics, mean has three related meanings)
[ link to this | view in chronology ]
That old parenting joke
[ link to this | view in chronology ]
Computers deal in facts, not political correctness. If a computer tells a researcher that his question is "Bullshit", chances are, the computer is the one who is right and not the researcher with hurt feelings.
I think this is something we'll have to come to terms with as a species at one point or another: sometimes reality just doesn't care if you get offended.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
If they train Watson as a diagnostic tool for veterinarians they can retain "bullshit".
[ link to this | view in chronology ]
I want to know what the researcher was querying.
[ link to this | view in chronology ]
Re: I want to know what the researcher was querying.
But that would imply that it wasn't actually profanity that the Semi-AI proclaimed but in fact the absolute truth.
But when has context ever gotten in the way of moralistic idiots who proclaim that profanity is somehow bad and uncivilised.
Chaucer must be rolling over in his grave now
[ link to this | view in chronology ]
"3.1415926535YoMama"
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: Re:
"It is 42 you fuck. lol"
[ link to this | view in chronology ]
Due to my profession, I find myself laughing that Watson, a computer intelligence, had a tiny bout of ADHD and no filters from going onto Urban Dictionary :-)
[ link to this | view in chronology ]
Jeopardy Champion
What the fuck is Fermat's Last Theorem.
[ link to this | view in chronology ]
Will Ferrell style rofl.
[ link to this | view in chronology ]
Re:
I'll take "Fuck You, Alex" for $1000.
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Lazy Scientists
Nice short term work around, but a real solution would seem to be right up their alley.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Submit your urban dictionary rejected words
[ link to this | view in chronology ]
[ link to this | view in chronology ]