| View previous topic :: View next topic |
| Author |
Message |
Carbon
Joined: 28 Jan 2011
|
|
| Back to top |
|
 |
Fat_Elvis
Joined: 17 Aug 2006
Location: In the ghetto
|
 Posted: Tue Feb 15, 2011 5:18 pm Post subject: |
 |
|
| I think these kind of developments will revolutionise TESOL. Watson is able to understand subtle nuances of language that were previously lost on artificial intelligence systems. I think we're not far off very accurate automatic translation systems. This will lead to a drop in demand for English as a lingua franca as there will be no demand for it. There will be a need to learn English in some cases, and in Korea social pressures to learn English are not necessarily related to any real need to learn English, but it will definitely change our industry. |
|
| Back to top |
|
|
Fat_Elvis
Joined: 17 Aug 2006
Location: In the ghetto
|
|
| Back to top |
|
|
Carbon
Joined: 28 Jan 2011
|
| Posted: Tue Feb 15, 2011 5:52 pm Post subject: |
|
|
Indeed. However, Watson (and the programmers) learns as it gains experience. It is clear that Watson will not be first, making the same mistake twice and second, given how often humans misunderstand one another, will soon exceed our accuracy of information.
The errors were quite interesting in themselves and are a very necessary part - as they are with humans - of learning. One of the "flubs" occurred because the computer cannot 'hear' the other answers and thus repeated the wrong answer, the other an error of omission; it actually had the right answer (it knew which body part), but with the format of the program, didn't formulate the complete question.
This idea of 'listening', while being quite a separate idea from the one at hand, is going to be the next hurdle in making this thing truly useful (in a broad, laymen sense of the word). I suspect that voice recognition will be added soon enough.
After Jeopardy:
http://www-943.ibm.com/innovation/us/watson/what-is-watson/watson-after-jeopardy.html
Again, it is interesting to think about how this could benefit learning language. |
|
| Back to top |
|
|
Carbon
Joined: 28 Jan 2011
|
|
| Back to top |
|
|
ddeubel
Joined: 20 Jul 2005
|
| Posted: Tue Feb 15, 2011 6:33 pm Post subject: |
|
|
Talk to alice - she's a trailblazer that I've added some English student friendly dialogue boxes to..... http://bit.ly/cInbAJ
this would take a longer discussion but if interested in the potential of technology and education - try English Central http://www.englishcentral.com - with your students. I'll also be looking for a lot more input from teachers soon - this can really help students. Make sure to sign up as a teacher so you can track students, assess them and guide their learning. It is perfect for Uni teachers...
Just loading up a video on EFL Classroom 2.0 on the original Turing test (Jeopardy is sort of a controlled variation). But the problem will never be meaning but rather our negotiation of that meaning - our forming of a social relationship through language when another party enters the fray. Colorless green ideas sleep furiously until another person enters the room.
It also begs the question as to whether computers can "think" rather than just answer.
DD
http://eflclassroom.com |
|
| Back to top |
|
|
Carbon
Joined: 28 Jan 2011
|
| Posted: Tue Feb 15, 2011 8:25 pm Post subject: |
|
|
| ddeubel wrote: |
Talk to alice - she's a trailblazer that I've added some English student friendly dialogue boxes to..... http://bit.ly/cInbAJ
this would take a longer discussion but if interested in the potential of technology and education - try English Central http://www.englishcentral.com - with your students. I'll also be looking for a lot more input from teachers soon - this can really help students. Make sure to sign up as a teacher so you can track students, assess them and guide their learning. It is perfect for Uni teachers...
Just loading up a video on EFL Classroom 2.0 on the original Turing test (Jeopardy is sort of a controlled variation). But the problem will never be meaning but rather our negotiation of that meaning - our forming of a social relationship through language when another party enters the fray. Colorless green ideas sleep furiously until another person enters the room.
It also begs the question as to whether computers can "think" rather than just answer.
|
What computers lack is emotion (I think this is where you are heading, ddeubel) and that is a huge factor in many human interactions and that is where computers will be hard pressed to catch humans. However, many scenarios don't involve emotion or would be better resolved without it interfering in what should be a clear decision or answer. Emotions may even be seen as the nemesis of thought. Could Watson console you if your puppy gets run over? Well, it could probably produce the right words for that situation (a very quantifiable scenario from a speech act point of view), but the question is would those words cheer us up as they would if they came from another person. Probably not, but that doesn't mean Watson would be somehow 'incorrect' in that situation. It would mean it isn't human, that's all.
Watson goes well beyond what an internet program can do. This is more than a program and clearly is not ready for language learning prime time. Until Watson can 'listen', it won't have the flexibility needed to be applicable across the many domains of ELL (or in other languages). There is already some talk of it being able to �read�.
Watch the last link to see exactly what is going on here. This is no keyword-based/ database referencing software, separating it from all we have known before. I understand the desire for Alice, but she is really just another deeply flawed program; a totally flawed version of what is really needed. Even though the software may be the best out there right now and indeed all that we have, it doesn't change how lame it is. English central is pretty much the same; a rich database of information, but again, relying on keywords and databases for interactivity.
Watson is negotiated meaning; the heart of it is to collect all manner and sources of information (non-selective) and produce consensus (answers). The database it uses is the same as ours; absolutely any source of text information. "Thinking" is a human term for an attribute; the word could easily be represented by another...'computing' perhaps. If most speech acts occur in quantifiable scenarios then they can be algorithmically predicted and the outcomes accurately represented or produced. These algorithms (processes) with the addition of a lifetime of experiences (the database, if you will) gives us conclusions or 'thoughts', which are then decoded in language. "Thinking" may then be a synthesis of information through quantifiable functions. Very do-able.
Rambling.
http://www.psychologytoday.com/blog/plato-pop/201102/watson-in-philosophical-jeopardy
http://www.pbs.org/wgbh/nova/tech/will-watson-win-jeopardy.html |
|
| Back to top |
|
|
RMNC
Joined: 21 Jul 2010
|
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
