The future of AI...

While I agree that this may be possible in theory, in practice it's not so easily done. First of all, consider the scale. The human brain has about 100 billion neurons, and each of them has weighted connections with thousands of others. Our technology just isn't mature enough to replicate such a mindnumbing piece of parallel computing. Maybe in a few decades, but it's impossible to tell.

Secondly, consider the fact that a brain is firmly entrenched in a body. It receives and sends messages from and the spine, organs, senses and what not. While all those things are closely researched, afaik, we are not at all close to understanding how all these things exactly interact with the brain. Even if we could replicate our bioware in hardware, but have no functional replica of the human body attached to it, it would be hard to predict what would happen if you would "throw on the switch", so to speak.

Thirdly, how do you "read" the human brain in order to make a copy in silicon hardware? Although we all share the same structures for vision, motion control, etc., our brains are very individual. You'd basically have to read it neuron for neuron, and follow the weighted connections. Given the numbers in my first paragraph, that's not only a daunting task, it's a near impossibility. Moreover, you'd also have to get the behaviour of each individual neuron right. They fire, or remain silent, based on their chemical make up. How do you read their chemistry? How do you examine their exact properties, in order to faithfully replicate them?

Why don't you give us your patent number, when all the paper work is done?

 

Log in or Sign up to hide all adverts.

Log in or Sign up to hide all adverts.

Kmguru:Your device seems like a very specialized example of an expert system.

Calling it an example of AI is a far cry from it actually being AI.

Deep Blue is not considered to be an AI device, and its activities seem far more complex than your system. It is merely a number cruncher.

BTW: 40 or more years ago, there were process control computers (and some very simple mechanical devices) which monitored the activities of complex chemical processes. A naive observer might consider such process control devices to be AI, but they did not come close to qualifying as such.

 

Log in or Sign up to hide all adverts.

True, Machines will use us.

 

Expecting us to develop psychokinetic abilites?

Perhaps you expect us to regress to a more primitive agricultural or hunter/gatherer culture.

Do you expect us to get stupid and develop some machine that we cannot turn off?

If I was part of a group which developed a machine whose capabilities were complex, potentially dangerous, and unpredictable, during the testing period (and maybe forever more), I would surely include some timing device to shut it off unless a member of the team reset the timer regularly.

 

Kmguru: I find it difficult to agree with any of your views on this issue.

What is the point of the following?

The above does not seem relevant to this thread or our posts to it. I said nothing about Bush or any entity unable to make intelligent decisions. Data is a fictional character. Discussion of god belongs in a forum dedicated to either religion or philosophy.

As of our current state of knowledge, I consider a human to be the only entity with general purpose intelligence. If data existed, I would consider him to be an AI, and would consider a much less capable robot to be an AI device.

I made some remarks about Deep Blue and process control computers/mechanisms, which I do not consider to be AI devices. I implied that the activities of Deep Blue are far more complex than your device. I also implied that the process control systems were comparable to your device. You replied with an appeal to authority, citing what was probably a publicity statement rather than an informed scientific opinion.

The above makes no attempt to compare the complexity of your device with Deep Blue or process control systems, neither of which are considered AI devices. Do you consider such devices to be AI? Do you consider your device to be more complex or closer to being an AI device?

You seem to me to be alone in your view. I know of nobody who considers highly specialized devices to be anything more than expert systems, rather than AI devices. Do you really consider your system to be more complex than Deep Blue, which (as I stated previously) is not even close to being AI.

I think that my view of what constitutes AI is closer to reality than yours.

I consider the above statement about my intelligence to be unbelievably arrogant if you are serious. At some time in the future, the human brain might be understood well enough to be described as a mere number cruncher. At present it is surely less well understood than even the most complex computer programs. Yet you claim to understand it well enough to be able to prove that one example of it is merely a number cruncher.

I can describe the basic design of Deep Blue, but defy anyone to give a similarly cogent description of how a human master plays chess. Even chess masters cannot give detailed cogent descriptions of how they make decisions. They make statements like: “My pieces were well developed and his position was cramped. An aggressive move was sure to be successful.” Even such statements might be after thoughts rather than opinions formed consciously at the time a critical move was made. I have never talked with a grand master, but I have discussed chess with players one or two levels down and way above my own ability (I am only a good coffeehouse player). Once an expert digresses from a book opening, his/her play seems to be more intuitive than rationally motivated. Of course their intuition has been honed by a thousand or more hours of play per year for many years. Yet you are claiming to be able to show that some particular human brain is merely a number cruncher.

Anybody who expects an exponential increase to continue indefinitely is a fool, especially if the exponent is as large as two.

I do not know who Ray is, but I find it hard to believe that he expects Moores’s law to continue to 2050 and beyond. Note that Moore’s law is merely an empirical observation, unlike a law such as Newton’s gravitational equations. I also believe that it was actually a statement about the density of electronic circuitry rather than computing power. Note that 50 years ago, a computer did simple instructions in about 2 milliseconds (500 per second). Doubling computing power every 18 months since then would result in current computers doing about 9*10⁸⁹ simple instructions per second. The fastest computer in about 1960 did about 500,000 operations per second (about 2 microseconds per operation). Moore’s law applied to the 1955 speed would predict about 900 million per second in 1960, and applying Moore’s law to the 1960 figure is silly (about 9*10¹⁷⁰).

I do not understand Ray’s remark about human computing power by 2050. Note that we probably have computers with more basic speed than a human brain today, and perhaps as much storage capacity. It is the mind’s organization that makes it better than modern computers. I see very little progress in computer architecture. Even the parallel processor systems with thousands of CPU’s have a primitive architecture compared to the human brain and do not come close to mimicking its general purpose intelligence.

As stated in a previous post, a highly specialized device is not considered to be an AI device by most people who think about AI.

 

I find it obvious that Dinosaur knows more about AI than you do, kmguru. Dinosaur seems to have an inkling of what AI actually is in the first place. You, on the other hand, seem to be building a wheel without a hub, referencing other researchers without discussing their actual work. Why will the pace of technological development increase exponentially until 2050, and what makes you so sure that it's exponentially increasing right now? Support your statements in your own words, please. Some of us don't have the time to read books as thick and dry as A New Kind of Science.

 

I honestly want you to at least make me consider the viability of your ideas. Are you refusing my challenge because you're incapable of supporting your argument, or are you just too lazy? You disappoint me in either case.

Would someone please stop this skirting of the issue and defend kmguru for him?

 

Well, thanks anyway. Maybe someone else who shares your position would be willing to try to explain it to me. If I wanted only to read about it, no doubt I would pick one of the books you mentioned above. I'm interested in discussing it, however, which is why I chose a message board as my medium of communication.

 

You could easily modify the problem by having the person in the room memorize the Chinese phrase book. Now the entire system consists of only the person, who can convincingly mimic understanding Chinese but is not actually able to do so.

 

Yes, my own quote, don't know what I was thinking. I meant of course, we will but won't be any more intelligent than us. Given enough time, I am pretty certain we could. But I don't think man will survive that long. He will be a victim of his own circumstances. But as computer science goes, one logic gate in itself can't be more complicated than itself unless networked with other logic gates, creating the ability to make more logic decisions. So when our brains, networked together as they are in society, come together, eventually we will make something more complicated than we are given enough time and enough people. So I guess I answered my own posted question right there. Now can we created a network of robots more complicated than our own network, or society? We would then need a network of more networks, so short answer, no.

 

Lol