Welcome to the North American Subaru Impreza Owners Club Friday March 29, 2024
Home Forums Images WikiNASIOC Products Store Modifications Upgrade Garage
NASIOC
Go Back   NASIOC > NASIOC Miscellaneous > Off-Topic

Welcome to NASIOC - The world's largest online community for Subaru enthusiasts!
Welcome to the NASIOC.com Subaru forum.

You are currently viewing our forum as a guest, which gives you limited access to view most discussions and access our other features. By joining our community, free of charge, you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content and access many other special features. Registration is free, fast and simple, so please join our community today!

If you have any problems with the registration process or your account login, please contact us.







* As an Amazon Associate I earn from qualifying purchases. 
* Registered users of the site do not see these ads. 
Reply
 
Thread Tools Display Modes
Old 05-18-2009, 10:50 AM   #1
mhoward1
Scooby Guru
 
Member#: 9481
Join Date: Aug 2001
Chapter/Region: South East
Location: FFR Challenge #43
Vehicle:
1832 Steam Buggy
Wood

Default so what ever happened to AI?

I remember seeing all these articles about Artificial Intelligence a few years ago and how we were so close. Whatever happened to all that research? Did it just die on the vine?
* Registered users of the site do not see these ads.
mhoward1 is offline   Reply With Quote
Sponsored Links
* Registered users of the site do not see these ads.
Old 05-18-2009, 10:58 AM   #2
El Camino
Scooby Newbie
 
Member#: 78760
Join Date: Jan 2005
Chapter/Region: MWSOC
Location: Open the door get on
Vehicle:
the floor everybody
walk the dinosaur

Default

Sarah Connor killed all of the developers.
El Camino is offline   Reply With Quote
Old 05-18-2009, 11:00 AM   #3
Jaredwhamilton
Scooby Newbie
 
Member#: 195550
Join Date: Nov 2008
Chapter/Region: MAIC
Location: Pennsylvania, United States
Vehicle:
2017 Subaru WRX STi
Crystal White Pearl

Default

^^

I haven't heard anything about it recently either.
Jaredwhamilton is offline   Reply With Quote
Old 05-18-2009, 11:00 AM   #4
mhoward1
Scooby Guru
 
Member#: 9481
Join Date: Aug 2001
Chapter/Region: South East
Location: FFR Challenge #43
Vehicle:
1832 Steam Buggy
Wood

Default

Quote:
Originally Posted by El Camino View Post
Sarah Connor killed all of the developers.
her or neo?
mhoward1 is offline   Reply With Quote
Old 05-18-2009, 11:01 AM   #5
Asinine
Sufficient for Sharky
 
Member#: 55173
Join Date: Feb 2004
Chapter/Region: MAIC
Location: I beat up big rocks
Vehicle:
with paddles.

Default

Yeah, I just google'd "artificial intelligence" and only got zero results
Asinine is offline   Reply With Quote
Old 05-18-2009, 11:01 AM   #6
cnw
Scooby Specialist
 
Member#: 98395
Join Date: Oct 2005
Default

it didn't do all that well at the box office iirc
cnw is offline   Reply With Quote
Old 05-18-2009, 11:03 AM   #7
Pain++
Scooby Specialist
 
Member#: 88289
Join Date: Jun 2005
Chapter/Region: Tri-State
Location: New York, NY
Vehicle:
2006 WRX: Stage 2
Urine-soaked Subway Car

Default

We had a pretty interesting discussion on it a few months back....

http://forums.nasioc.com/forums/show...l+intelligence
Pain++ is offline   Reply With Quote
Old 05-18-2009, 11:19 AM   #8
AWDshenanigans
Scooby Specialist
 
Member#: 197856
Join Date: Dec 2008
Chapter/Region: SWIC
Location: 'merikuh
Vehicle:
2010 FJ Cruiser
Sandstorm

Default

its not within the grasp of current computers and probably wont be for decades. computers are better than humans at storing data and processing it, but they cant understand what they are computing. AI is dependant on your individual definition of AI. i personally think a computer would have to have an opinion on something for me to call it AI, some people have other benchmarks. I also think its physically impossible for computers as they are built today to acheive AI. think about it, the "knowledge" of your computer is all processed as binary code (one's and zero's for...well, sorry you shouldnt be talking about AI if you dont know what binary is). the computer stores all the data in binary so if you were to have a REALLY big stack of paper you could theoretically write out everything a computer "knows". (i didnt take any steps to verify this, but i was recently told the bed of an average truck could be filled with sheets of paper to store roughly one gig of info) anyway, this fact makes artificial intelligence based on computers of today impossible because no matter how powerful they are, they're just a glorified list. so if a computer as we know it today can achieve AI, then so can a giant stack of paper.
AWDshenanigans is offline   Reply With Quote
Old 05-18-2009, 11:30 AM   #9
Physics Junkie
Scooby Specialist
 
Member#: 65609
Join Date: Jul 2004
Chapter/Region: SCIC
Location: <3s me some SCIC.
Default

Weak AI or strong AI? There is a ton of crap going on in statistical learning, computer vision, etc. You would have to be more specific and probably want to ask elsewhere. The popular notion of creating something that responds to a turing test is dull and lacking in imagination. No, I don't want a robot maid a la the Jetsons. I want Google to map the internets more efficiently.

My uneducated and humble guess is that we'll get significantly more out of giving normal humans neural implants than we will out of a computer trained to behave like a computer. I volunteered in this dude's lab back in college: http://www.neural-prosthesis.com/ so that sways my opinion.
Physics Junkie is offline   Reply With Quote
Old 05-18-2009, 12:23 PM   #10
karnick
Scooby Newbie
 
Member#: 103863
Join Date: Dec 2005
Chapter/Region: Tri-State
Location: Highlands, NJ
Default

The problem is that it's hard to find a suitable model for learning just anything. Neural networks are used to implement machine learning, but what's being learned is pretty restricted. Most of the AI work I have done involves pattern recognition, and automated decision making. There are a LOT of obstacles when it comes to learning though. You always need to know SOMETHING about what you are learning, which in a way can be thought of as 'cheating' since you are giving the model a head start. In research these are just considered 'assumptions'. Also, what if what you are learning changes over time? Learning what the letter B looks like is different than learning how a stock will react. It's hard to find one model to do everything.

If you are seriously interested in seeing exactly where we are with AI research you should browse conference and journal papers through IEEE or the ACM. You can google certain conferences, such as IJCNN, and there will be some short 6 page papers on various topics. I've been published with 3 'machine learning' papers, one on diagnosing alzheimers disease, and two on learning concept drift (similar to the stock problem). It is an extremely difficult research area because everything is experimental and there are not many standards to base your results on. You basically have to justify them yourself and prove they are relevant.


On a more general note, computing power is also a huge obstacle when trying to perform a large number of varied experiments in a reasonable amount of time.
karnick is offline   Reply With Quote
Old 05-18-2009, 12:28 PM   #11
BGPunk2001
Scooby Specialist
 
Member#: 46134
Join Date: Oct 2003
Chapter/Region: TXIC
Location: Ignoranimosity
Default

Quote:
Originally Posted by karnick View Post
The problem is that it's hard to find a suitable model for learning just anything. Neural networks are used to implement machine learning, but what's being learned is pretty restricted. Most of the AI work I have done involves pattern recognition, and automated decision making. There are a LOT of obstacles when it comes to learning though. You always need to know SOMETHING about what you are learning, which in a way can be thought of as 'cheating' since you are giving the model a head start. In research these are just considered 'assumptions'. Also, what if what you are learning changes over time? Learning what the letter B looks like is different than learning how a stock will react. It's hard to find one model to do everything.

If you are seriously interested in seeing exactly where we are with AI research you should browse conference and journal papers through IEEE or the ACM. You can google certain conferences, such as IJCNN, and there will be some short 6 page papers on various topics. I've been published with 3 'machine learning' papers, one on diagnosing alzheimers disease, and two on learning concept drift (similar to the stock problem). It is an extremely difficult research area because everything is experimental and there are not many standards to base your results on. You basically have to justify them yourself and prove they are relevant.


On a more general note, computing power is also a huge obstacle when trying to perform a large number of varied experiments in a reasonable amount of time.
What about when people are given something new to learn and they try to relate it to some other thing that they have experienced before. That is what I generally do when learning something new. It helps me understand it and you can't really consider that cheating.

You are just trying to give the new lesson context.
BGPunk2001 is offline   Reply With Quote
Old 05-18-2009, 12:30 PM   #12
Fuzz541
Scooby Specialist
 
Member#: 4739
Join Date: Mar 2001
Chapter/Region: NESIC
Location: Jiangxi, China
Default

It left and now hangs out mostly on teh b4rn and gunatics.
Fuzz541 is offline   Reply With Quote
Old 05-18-2009, 12:32 PM   #13
aod
Scooby Newbie
 
Member#: 7121
Join Date: Jun 2001
Chapter/Region: MWSOC
Location: D2F.1 = D2F.2, D2F.3 = D2F.4
Default

There is a ton of things going on with AI, it just happens behind the scenes. Ever wonder how your TiVO automagically starts recording stuff you like to watch? Voice recognition has come leaps and bounds from where it was just a few years ago. Automation is constantly pushing forward. Those are just a few examples.
aod is offline   Reply With Quote
Old 05-18-2009, 12:34 PM   #14
karnick
Scooby Newbie
 
Member#: 103863
Join Date: Dec 2005
Chapter/Region: Tri-State
Location: Highlands, NJ
Default

Quote:
Originally Posted by BGPunk2001 View Post
What about when people are given something new to learn and they try to relate it to some other thing that they have experienced before. That is what I generally do when learning something new. It helps me understand it and you can't really consider that cheating.

You are just trying to give the new lesson context.
I probably should have given an example. The way I explained it can probably be interpreted in a lot of different ways.

Let's say you are trying to have a computer learn the difference between Alzheimer's disease and non-Alzheimer's disease by feeding it EEG signals (brain waves). You are restricting what the model can actually learn. A model can't just start learning 'anything' and give feedback, you have to narrow the problem down to something more specific.

Basically, the AI model has to know what you are teaching it, and it's acquired knowledge is limited to this. That's what I meant. Hope that helps.
karnick is offline   Reply With Quote
Old 05-18-2009, 12:34 PM   #15
karnick
Scooby Newbie
 
Member#: 103863
Join Date: Dec 2005
Chapter/Region: Tri-State
Location: Highlands, NJ
Default

Quote:
Originally Posted by aod View Post
There is a ton of things going on with AI, it just happens behind the scenes. Ever wonder how your TiVO automagically starts recording stuff you like to watch? Voice recognition has come leaps and bounds from where it was just a few years ago. Automation is constantly pushing forward. Those are just a few examples.
This. The TiVO example is one of those AI problems you can think of with a lot of assumptions/rules.
karnick is offline   Reply With Quote
Old 05-18-2009, 12:35 PM   #16
.brian.
Scooby Newbie
 
Member#: 155207
Join Date: Jul 2007
Chapter/Region: Tri-State
Location: Bad things happen in Philly
Vehicle:
2020 MDX
2022 Odyssey

Default

he got traded to the pistons.

"practice? practice? we talkin' bout practice?"
.brian. is offline   Reply With Quote
Old 05-18-2009, 12:43 PM   #17
BGPunk2001
Scooby Specialist
 
Member#: 46134
Join Date: Oct 2003
Chapter/Region: TXIC
Location: Ignoranimosity
Default

Quote:
Originally Posted by karnick View Post
I probably should have given an example. The way I explained it can probably be interpreted in a lot of different ways.

Let's say you are trying to have a computer learn the difference between Alzheimer's disease and non-Alzheimer's disease by feeding it EEG signals (brain waves). You are restricting what the model can actually learn. A model can't just start learning 'anything' and give feedback, you have to narrow the problem down to something more specific.

Basically, the AI model has to know what you are teaching it, and it's acquired knowledge is limited to this. That's what I meant. Hope that helps.
Well, yeah, but basically that's what we do right? I wouldn't know the difference between those two either.


Lets say I had never seen or heard of an apple. If you gave me one and asked me what it was I would tell you I had no clue.


Now, lets say I have never seen an apple in person, but it has been described to me before. Now, I actually have a decent chance at guessing what it was. But someone had to tell me at one point.


Is that not somewhat the same scenario you are describing, or am I way off base?
BGPunk2001 is offline   Reply With Quote
Old 05-18-2009, 12:47 PM   #18
HoleSh0t
Scooby Newbie
 
Member#: 64007
Join Date: Jun 2004
Chapter/Region: MWSOC
Location: Page 2
Vehicle:
2004 Nissan 350Z
XBL: Hollowp0int

Default

HoleSh0t is offline   Reply With Quote
Old 05-18-2009, 01:04 PM   #19
karnick
Scooby Newbie
 
Member#: 103863
Join Date: Dec 2005
Chapter/Region: Tri-State
Location: Highlands, NJ
Default

Quote:
Originally Posted by BGPunk2001 View Post
Well, yeah, but basically that's what we do right? I wouldn't know the difference between those two either.


Lets say I had never seen or heard of an apple. If you gave me one and asked me what it was I would tell you I had no clue.


Now, lets say I have never seen an apple in person, but it has been described to me before. Now, I actually have a decent chance at guessing what it was. But someone had to tell me at one point.


Is that not somewhat the same scenario you are describing, or am I way off base?
You're on the right track. Think of the brain as one general model composed of other smaller models. The thing is, the brain is powerful enough to see this new apple and create a new model on the fly and adjust it over time. With AI, it is hard to just create a brand new model for something such as an apple that has never been seen before without manually providing it some underlying information. The brain is far more complex than any model we can generate. It's hard to create an algorithm that can learn something new and differentiate it from other things with no prior knowledge.
karnick is offline   Reply With Quote
Old 05-18-2009, 01:26 PM   #20
AWDshenanigans
Scooby Specialist
 
Member#: 197856
Join Date: Dec 2008
Chapter/Region: SWIC
Location: 'merikuh
Vehicle:
2010 FJ Cruiser
Sandstorm

Default

Quote:
Originally Posted by aod View Post
There is a ton of things going on with AI, it just happens behind the scenes. Ever wonder how your TiVO automagically starts recording stuff you like to watch? Voice recognition has come leaps and bounds from where it was just a few years ago. Automation is constantly pushing forward. Those are just a few examples.
tivo works by comparing lists and doing calculations. its a glorified flow chart. same with voice recognition.
AWDshenanigans is offline   Reply With Quote
Old 05-18-2009, 01:31 PM   #21
Mad06STi
Scooby Newbie
 
Member#: 123324
Join Date: Aug 2006
Chapter/Region: MWSOC
Location: Madison, WI
Vehicle:
2006 STi
07 Daytona 675

Default

Quote:
Originally Posted by AWDshenanigans View Post
tivo works by comparing lists and doing calculations. its a glorified flow chart. same with voice recognition.
As already stated in this thread and every thread about AI, it depends on how you define AI. Most AI today is some type of classifier (a glorified flow chart), but it is still pretty advanced.
Mad06STi is offline   Reply With Quote
Old 05-18-2009, 01:32 PM   #22
varmint007
Scooby Newbie
 
Member#: 13626
Join Date: Dec 2001
Chapter/Region: MAIC
Location: Maryland
Vehicle:
02 996

Default

This Guy Jeff Hawkins started the company Palm
http://en.wikipedia.org/wiki/Jeff_Hawkins

Now he has another company which is supposed to be the practical application of a modernized view of artificial intelligence:
http://www.numenta.com/

The model is basically that the brain is constantly predicting what will happen (based on prior experiences) then comparing that prediction to the actual results. Numenta makes and gives away software that is designed to mimic this model of the brain.

His point is that having computers go FASTER is meaningless if its programmed to qualitatively perform the wrong calculations.

not sure if they have any good results yet - it's been a while since I was following it

here's his book describing the model:
http://www.amazon.com/Intelligence-Jeff-Hawkins/dp/0805078533/ref=pd_lpo_k2_dp_k2a_3_txt?pf_rd_p=304485601&pf_rd_s=lpo-top-stripe-2&pf_rd_t=201&pf_rd_i=0805074562&pf_rd_m=ATVPDKIKX0DER&pf_rd_r=03GVC78J2QD58NQ23R8H
varmint007 is offline   Reply With Quote
Old 05-18-2009, 01:42 PM   #23
Dave D.
NASIOC Supporter
 
Member#: 130990
Join Date: Nov 2006
Chapter/Region: NWIC
Location: Menzel Lake, WA
Vehicle:
06 Red Ltd.Wagon
Dom 1.5xtr JJ tuned

Default

Scientists are just now coming out with models that show the physical brain just the tip of the iceberg; the real brain operations occur in the realm of quantum physics; infinitely more complicated and much tougher to duplicate.

Quantum computing >>> Brute force computing.
Dave D. is offline   Reply With Quote
Old 05-18-2009, 01:45 PM   #24
AWDshenanigans
Scooby Specialist
 
Member#: 197856
Join Date: Dec 2008
Chapter/Region: SWIC
Location: 'merikuh
Vehicle:
2010 FJ Cruiser
Sandstorm

Default

Quote:
Originally Posted by Mad06STi View Post
As already stated in this thread and every thread about AI, it depends on how you define AI. Most AI today is some type of classifier (a glorified flow chart), but it is still pretty advanced.
i was the one that stated it. and my definition is a computer doesnt have AI unless it can have an opinion on something.
AWDshenanigans is offline   Reply With Quote
Old 05-18-2009, 01:51 PM   #25
Mad06STi
Scooby Newbie
 
Member#: 123324
Join Date: Aug 2006
Chapter/Region: MWSOC
Location: Madison, WI
Vehicle:
2006 STi
07 Daytona 675

Default

Quote:
Originally Posted by AWDshenanigans View Post
i was the one that stated it. and my definition is a computer doesnt have AI unless it can have an opinion on something.
And I would agree, very little in the AI world interests me at this point.
Mad06STi is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
so, what ever happened to ... SQ3.0dotJP Off-Topic 54 03-15-2009 09:08 PM
So what ever happened to the free Driving Sport CD for current subscribers? Geese1 Grassroots Motorsports Magazine Forum 2 08-17-2006 11:36 AM
so what ever happened to padre [27] Texas Impreza Club Forum -- TXIC 7 10-19-2002 10:28 AM
So what ever happened to AZScooby logo? olsonjj South West Impreza Club Forum -- SWIC 9 04-16-2001 12:58 AM

All times are GMT -4. The time now is 04:35 AM.


Powered by vBulletin® Version 3.7.0
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Powered by Searchlight © 2024 Axivo Inc.
Copyright ©1999 - 2019, North American Subaru Impreza Owners Club, Inc.

As an Amazon Associate I earn from qualifying purchases.

When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission
Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.