Artificial Intelligence Depot
Visiting guest. Why not sign in?
News, knowledge and discussion for the AI enthusiast.
FEATURES COMMUNITY KNOWLEDGE SEARCH  
The Intelligent Wireless Web
A vision of the Web's near future and overviews the technologies that will make it possible, exploring developments in speech recognition, mobile wireless devices, network integration, and software development.
More information at Amazon US UK

Reply to Message

Not registered yet?

The AI Depot has a focused community of friendly users. Rather than let anyone abuse the site at the brink of promiscuity, we prefer to let only those with an active interest participate... this simply requires registering.

Why not sign up!

Joining the site's community is completely free. You can then post messages freely, and customise your personal profile at will. Specific privileges will also be granted to you, like being able to access printer-friendly articles without restrictions. So, why not register?

Username:
Password:
Subject:
Email me when someone replies.
Body:

Parent Message

Whither stong AI?

It depends on what you think of as "strong AI". If you think of it as a sort of robotic butler depicted in asimov stories then there are quite a few practical engineering issues which stand in the way of building such a thing. We have man-sized robots that can walk but they're extremely expensive and their batteries typically only last for 20 minutes before they need recharging. Lightweight materials and actuators need to be developed which could power a man-sized machine and enable it to carry out useful work for a sustained period of time.

But the mechanical engineering problems are relatively straightforward compared to the problems of cognition. There still isn't really any coherent theoretical framework which describes in detail how the brain works as a system. We have a few techniques (such as neural networks, genetic algorithms and so on), and some ideas taken from biology, psychology and ethology but so far these give only a highly fragmented view of how the system works overall. Probably the closest that anyone has come to an overall theory of brain function was Gerald Edelman's "neural darwinism", but even this has limited explanetory power I think.

Nevertheless I'm fairly optomistic that we do have sufficient technology and understanting to produce human-like machines which could carry out some types of manual work, and interact with people in a limited (perhaps childlike) way in the near future. Imagine an ASIMO robot with a Kismet-like head which could behave with a limited degree of autonomy and also be capable of teleoperation via a wireless connection to the internet. The only real question is whether this could be achieved at a cost which was suitable for the mass market.

- Bob

136 posts.
Tuesday 16 July, 17:48
Reply
Well said, Bob

I agree with Bob: what's fundamentally lacking from 'human-like' artificial intelligence is the kind of associative, and pseudo-creative algorithms with which we conduct our daily lives. To say that reproducing human intelligence in a machine is the ultimate goal, however, is a misguided notion.

Human intelligence is simply a fancy sort of Basic Input Output System, and reproducing a higher form of intelligence would therefore be a far greater accomplishment. There is nothing special, spiritual or divine about our brains - we merely learn the most productive way to respond to our senses and emotions over time. The seed for OUR development is the survival instinct, however, we now find ourselves in a position to provide whatever seed we choose to machines so that they may evolve in a completely different (not lesser, equal or better) way, towards any goal imaginable.

For the time being though, even if we had cracked the biological 'life' algorithm (no doubt used in our own creation), the kind of digital technology to which we would apply it would yield results vastly different from what we might have expected.

2 posts.
Thursday 18 July, 12:12
Reply
The naked ape

Ever since the time of AM Turing people have aspired to machines of human-level intelligence, and I suspect that nothing has really changed in that respect.

I wouldn't agree with your comments about human intelligence being "simply" I/O or "merely" learning and survival instincts. Actually the more I read about these things and actually try to produce some simulations of them the more respect I have for the the sheer complexity and the dizzying levels of sophistication of the human brain compared to the rather crude technologies and methods that we have at the moment. None of the AI systems which anyone has come up with yet get even close to its adaptibility and capacity for imaginative invention.

The ultimate goal for people like Kurzweil or Moravec would be to reach human-level intelligence and then race beyond it into some sort of weird transhuman future. For some of these guys technology seems to have been elevated almost to the status of a religeon. Maybe these sorts of transcendent fantasy could be possible one day, but I think that sort of thing is so far away into the future that any speculations made about it are almost certain to turn out wrong.

- Bob

136 posts.
Thursday 18 July, 17:52
Reply
Smart now, but ....

Okay, human intelligence may seem sophisticated now, but what I was referring to was the extremely simplistic "I/O" used in its evolution from single-celled organisms. Every facet of our intelligence, indeed our psyche, is undeniably an offshoot of our survival instict, whatever way you choose to look at it.

2 posts.
Friday 19 July, 04:19
Reply
tabletop and Dr. Robert French

Dr. French wrote a program called tabletop while he was working at Indiana with Doug Hofstaedter. He documented it in a book called "The Subtelty of Sameness" which focuses on analogy making as part of what makes us human. I think tabletop is a very good approach to modeling human thought, but the program could not learn from its mistakes. It would be interesting to see what would happen if learning were incorporated into it.

I personally think either a Rod Brooks or Doug Hofstaedter approach will end up solving the biggest AI problems we currently face.

Rob

15 posts.
Friday 19 July, 09:57
Reply

Back to the Artificial Intelligence Depot.