Have you guys seen this?
I missed it last nite, need to set the DVR. Looks super cool!
http://gizmodo.com/#!5760858/whats-next-for-the-jeopardy-computer





Have you guys seen this?
I missed it last nite, need to set the DVR. Looks super cool!
http://gizmodo.com/#!5760858/whats-next-for-the-jeopardy-computer
http://gizmodo.com/#!5760514/ibm-jeopardy-challenge-night-1-watson-ties-for-the-lead
We DVR'ed it and watched it after coming home from the festivities.
First thoughts: Ken and Brad seemed to change their approach as the game progressed. At the beginning, it seemed like they would figure out the question first, then ring the buzzer. But since Watson can process then buzz much faster than K&B, the humans seemed to catch on and started buzzing in even before they knew the answer, buying them a couple more seconds to figure out the question between the buzz and answering Trebek.
Also, really interesting how Watson chooses his next question. I think Brad only tied because he went straight for the $1000 answers.
shmack wrote >>
We DVR'ed it and watched it after coming home from the festivities.
First thoughts: Ken and Brad seemed to change their approach as the game progressed. At the beginning, it seemed like they would figure out the question first, then ring the buzzer. But since Watson can process then buzz much faster than K&B, the humans seemed to catch on and started buzzing in even before they knew the answer, buying them a couple more seconds to figure out the question between the buzz and answering Trebek.
Also, really interesting how Watson chooses his next question. I think Brad only tied because he went straight for the $1000 answers.
I caught the same strategy. Assuming the machine cannot alter its basic algorithm to match, it is going to be close.
I also note that the IBM crew used a voice that is just WAY too similar to the HAL 9000. I half expect to hear "I'm sorry Alex, I'm afraid I can't do that". I also wonder if he'll try to open the airlocks and kill his opponents if he thinks the mission is 'in jeopardy'.
Rockmastermike wrote >>
shmack wrote >>
We DVR'ed it and watched it after coming home from the festivities.
First thoughts: Ken and Brad seemed to change their approach as the game progressed. At the beginning, it seemed like they would figure out the question first, then ring the buzzer. But since Watson can process then buzz much faster than K&B, the humans seemed to catch on and started buzzing in even before they knew the answer, buying them a couple more seconds to figure out the question between the buzz and answering Trebek.
Also, really interesting how Watson chooses his next question. I think Brad only tied because he went straight for the $1000 answers.I caught the same strategy. Assuming the machine cannot alter its basic algorithm to match, it is going to be close.
I also note that the IBM crew used a voice that is just WAY too similar to the HAL 9000. I half expect to hear "I'm sorry Alex, I'm afraid I can't do that". I also wonder if he'll try to open the airlocks and kill his opponents if he thinks the mission is 'in jeopardy'.
Freaky!
I noticed the strategy shift too. It also seemed that Watson's "confidence" went down once we got into the $1000 questions, so it will be interesting to watch double jeopardy tonight.
I still can't decide if I'm rooting for skynet or for ken yet.
I was fascinated by this. Must see TV IMO.
You know we're going to have to go back in time and keep this from happening, right?
Rockmastermike wrote >>
shmack wrote >>
We DVR'ed it and watched it after coming home from the festivities.
First thoughts: Ken and Brad seemed to change their approach as the game progressed. At the beginning, it seemed like they would figure out the question first, then ring the buzzer. But since Watson can process then buzz much faster than K&B, the humans seemed to catch on and started buzzing in even before they knew the answer, buying them a couple more seconds to figure out the question between the buzz and answering Trebek.
Also, really interesting how Watson chooses his next question. I think Brad only tied because he went straight for the $1000 answers.I caught the same strategy. Assuming the machine cannot alter its basic algorithm to match, it is going to be close.
I also note that the IBM crew used a voice that is just WAY too similar to the HAL 9000. I half expect to hear "I'm sorry Alex, I'm afraid I can't do that". I also wonder if he'll try to open the airlocks and kill his opponents if he thinks the mission is 'in jeopardy'.
Thinking the same thing about the HAL voice; would think they would shy away from that just from the connotations. It was quite interesting to watch and will have to go back and watch that episode of Nova on it now.
Kill All Humans for 1600
Let's just hope our microchip-based overlords are armed with Windows Vista. We'd have a hope in that case.
This was on last friday...
http://www.pbs.org/wgbh/nova/tech/smartest-machine-on-earth.html
As I developer I thought it was really interesting how they used duck typing to create a sort of inference grouping. Then added rules to "wire" everything together.
I would be really interested in seeing this technology facilitate more financial regulation, by breaking through the complex math of things like derivatives. Preventing banks from constructing inflated capital reserves.
GViewProgrammer wrote >>
I would be really interested in seeing this technology facilitate more financial regulation, by breaking through the complex math of things like derivatives. Preventing banks from constructing inflated capital reserves.
even more interesting would be breaking through the massive money wall of lobbyists preventing any kind of meaningful reregulation...
that may be beyond modern computer science for quite some time.
(But I do like where that idea was going)
You must log in to post.