U.S. CitiesCity-Data Forum Index
Go Back   City-Data Forum > General Forums > Economics
 [Register]
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
View detailed profile (Advanced) or search
site with Google Custom Search

Search Forums  (Advanced)
Reply Start New Thread
 
Old 05-23-2018, 11:35 AM
 
5,603 posts, read 4,214,284 times
Reputation: 10567

Advertisements

Quote:
Originally Posted by rruff View Post
...... Give it a little time...

If you are talking about AI, I have been. As I mentioned I saw my first attempt to utilize AI 40 years ago.
Reply With Quote Quick reply to this message

 
Old 05-23-2018, 12:12 PM
 
Location: Ruidoso, NM
5,170 posts, read 4,738,304 times
Reputation: 4206
Quote:
Originally Posted by jrkliny View Post
If you are talking about AI, I have been. As I mentioned I saw my first attempt to utilize AI 40 years ago.
Yes, with the processing power of... well very little.

As the processing power becomes powerful enough and cheap enough, do you really believe algorithms will be an insurmountable barrier? What are the algorithms that govern human and animal functioning (honed over billions of generations of natural selection)? Why would you assume that AI could not similarly learn? You seem to focus only on the areas where it has failed to match humans... with hardware that is very subhuman.
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 01:11 PM
 
Location: Omaha, Nebraska
6,312 posts, read 3,490,581 times
Reputation: 14999
Quote:
Originally Posted by rruff View Post
As the processing power becomes powerful enough and cheap enough, do you really believe algorithms will be an insurmountable barrier? What are the algorithms that govern human and animal functioning (honed over billions of generations of natural selection)?
That's a big part of the problem: we literally don't know. And because we don't know, we can't replicate what the human brain is actually doing when it learns how to think.
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 01:29 PM
 
Location: Ruidoso, NM
5,170 posts, read 4,738,304 times
Reputation: 4206
Quote:
Originally Posted by Aredhel View Post
That's a big part of the problem: we literally don't know. And because we don't know, we can't replicate what the human brain is actually doing when it learns how to think.
We don't need to replicate it. We know that animals (including humans) have pre-programmed and learned responses to their environment and methods of achieving goals. AI will similarly produce solutions in its own way, and it many cases (not all) it can already far exceed human capability.

Read up on the AI that learned to play Go. No humans can fathom how it "thinks".
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 01:45 PM
 
5,603 posts, read 4,214,284 times
Reputation: 10567
Again, for a great many applications the limitations are not computing power. The algorithms are the issue. For the exciting future applications there is no hope in ell that a human could develop them. We have proven those limitations on relatively simple tasks in the past. AI and significant future applications means that the machines need to develop their own algorithms. With neural networks and similar approaches they do this largely by trial and error. The machine builds algorithms and selects and modifies those algorithms based on feedback. The feedback is not some internal computer process. It comes from the outside and that is a major limitation that has nothing to do with computing power. There is another even more basic issue. A computer is not some sort of self learner as are humans and other animals. A computer needs to be programmed as to how to learn and what to learn. That is another serious weak link that depends entirely on us feeble humans. We barely understand how computers operate and have virtually no understanding of how we operate. We are no where near ready to understand human cognition, sensory and brain functions and no where near being able to transfer that sort of capability to a computer. How do you tell a computer what it should and what it needs to learn?


A computer cannot in any fashion duplicate human learning. Look at how a baby learns. First the baby is prewired and preprogrammed to utilize the capability of our senses. Our whole brains and nervous systems are structured to utilize and coordinate sensory inputs. It is not just the physical structure, but we know that the "algorithms" that are innate are very sophisticated. We also have additional characteristics completely missing from machines. We are born with a level of curiosity and drive to learn and to explore our worlds. Even more importantly we have feelings and emotions that not only drive us but are affected by the feedback from the environment. As a baby's brain and senses grow and mature, the individual is constantly learning by interacting with the environment. Some of the learning is directed by parents but the vast majority is non-structured. It is constant touching, tasting, pushing, throwing, and countless other forms of interaction with the environment. It goes on minute by minute, hour by hour, day by day and for years. In fact indefinitely.


Computers are powerful in doing some things. We have been taking advantage of those powerful characteristics. Learning is not some thing they do well. We barely understand how to have a computer do even the most primitive and directed learning. Having a massively huge supercomputer that has learned how to play a board game is really not that impressive. Humans were able to program deep blue to play chess years ago. When it comes to developing computers that can learn we are not near the human level. Not near the level of a slug. We are down there at the level of the most primitive organisms. We have actually been there for quite a while and huge advances in computational power have not seemed to make much difference.
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 07:03 PM
 
24,956 posts, read 11,634,183 times
Reputation: 11663
Quote:
Originally Posted by jrkliny View Post
A computer cannot in any fashion duplicate human learning. Look at how a baby learns. First the baby is prewired and preprogrammed to utilize the capability of our senses. Our whole brains and nervous systems are structured to utilize and coordinate sensory inputs.

that was a huge wall of text to get to this basic concept that you have that you use to justify it.


We are literally teaching computers. They ARE learning like us. You have this idea that we are trying to create AI's by writing code, but really what we are doing is writing code to create that prewired state. The newer AI's are learning by being shown things. No one is trying to write code to create a AI, instead we're writing code that creates a internal infrastructure that allows a computer to learn as we do.


AI is not being developed the way you are arguing. And its getting better at a massive rate.
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 07:08 PM
 
24,956 posts, read 11,634,183 times
Reputation: 11663
Quote:
Originally Posted by jrkliny View Post
If you are talking about AI, I have been. As I mentioned I saw my first attempt to utilize AI 40 years ago.

yeah me too. And 40 years ago it wasnt going anywhere. It was pie in the sky theories and experiments that did nothing useful. You probably know Lisp if you were messing around with it back then.



Today is different. We have low level working systems. Things that can identify objects with incomplete data. Things that learn.
Reply With Quote Quick reply to this message
 
Old 05-23-2018, 09:44 PM
 
5,603 posts, read 4,214,284 times
Reputation: 10567
Quote:
Originally Posted by greywar View Post
......
...... You have this idea that we are trying to create AI's by writing code, but really what we are doing is writing code to create that prewired state. The newer AI's are learning by being shown things. .......
The code by which an AI computer learns has to be written. The machine does not create it out of air. The current processes are at a primitive state.


AI does indeed learn, but again the process requires feedback. I will again use a self driving car as an example since that seems to be one of the main topics of concern. Some of the algorithms have been written by humans. Most are learned through a trial and error process that is slow and obviously has not worked well. The issue is not the computing power but the software by which the computer operates and learns.
Reply With Quote Quick reply to this message
 
Old 05-24-2018, 12:30 PM
 
Location: Honolulu, HI
4,636 posts, read 1,156,512 times
Reputation: 6627
AI is not close to doing anything the Op describe nor will it “generate free money” from the stock market. Capitalism, supply and demand, nor the stock market works like that.

Everyone can’t be a “winner with endless wealth machines.”

People are so obsessed with AI trying to be some socialist gap closer they forget to turn on their brain and realize how the economy/stock market will immediately adjust to it. These “ AI money machines by the masses” would immeditely cause hyperinflation which means a gallon of milk would cost $10,000
Reply With Quote Quick reply to this message
 
Old 05-24-2018, 01:17 PM
Status: "If you like healthcare, you'd better vote Democrat" (set 22 days ago)
 
Location: Atlanta, GA
181 posts, read 74,995 times
Reputation: 425
Quote:
Originally Posted by Rocko20 View Post
AI is not close to doing anything the Op describe nor will it “generate free money” from the stock market. Capitalism, supply and demand, nor the stock market works like that.

Everyone can’t be a “winner with endless wealth machines.”

People are so obsessed with AI trying to be some socialist gap closer they forget to turn on their brain and realize how the economy/stock market will immediately adjust to it. These “ AI money machines by the masses” would immeditely cause hyperinflation which means a gallon of milk would cost $10,000
Sigh. You just can't win, no matter how hard you try.

But I'm not one to give up quite so easily. Who needs the stock market when I can have my AI set up 3 or 4, or how about 10 avatars who can do really high-skilled jobs and get paid $$$ for them? That's creating real wealth, right?

If that fails, I'll try something else. One thing's for sure, having my very own personal AI is gonna be quite the adventure.
Reply With Quote Quick reply to this message
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.

Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.


Reply
Please update this thread with any new information or opinions. This open thread is still read by thousands of people, so we encourage all additional points of view.

Quick Reply
Message:

Over $104,000 in prizes was already given out to active posters on our forum and additional giveaways are planned!

Go Back   City-Data Forum > General Forums > Economics
Follow City-Data.com founder on our Forum or

All times are GMT -6.

2005-2018, Advameg, Inc.

City-Data.com - Archive 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35 - Top