Ethics & destiny of AI creations

[shunyadragon]

Prediction of England hurricane

26 October 2013 - The Met Office is warning of the risk of a significant storm bringing exceptionally strong winds to parts of England and Wales on Sunday night into Monday morning.

Currently forecasts suggest a low pressure system will rapidly deepen just to the south west of the UK on later on Sunday, before moving across the country to be out over the North Sea by the afternoon on Monday.

This is expected to bring gusts of 60 - 80 mph widely across the southern half of the UK, with gusts of more than 80 mph possible in places - especially on exposed coasts.

Any major storm which occurs in early autumn has the potential to cause widespread severe disruption through falling trees, structural damage, transport disruption or power cuts and possibly flooding.

Frank Saunders, Chief Forecaster at the Met Office, said: "We are confident that a severe storm will affect Britain on Sunday night and Monday. We are now looking at refining the details about which areas will see the strongest winds and the heaviest rain.

[Sparko]

Prediction of England hurricane

Yeah but forecasts are done based on observations and satellite imagery, not computer modeling (other than predicting possible paths of storms which are often wrong). notice how it speaks of "risk" and "potential?" --- that's because it is not being accurately modeled but merely forecast by observation.

When we are talking modeling we mean being able to model the weather for the planet, and accurately predict future storms and weather patterns well before they occur. We can't do that. We probably never will be able to do that. Small initial variables can result in wildly different outcomes. And there is no way to know what all of those initial variables are.

[shunyadragon]

Yeah but forecasts are done based on observations and satellite imagery, not computer modeling (other than predicting possible paths of storms which are often wrong). notice how it speaks of "risk" and "potential?" --- that's because it is not being accurately modeled but merely forecast by observation.

False once the storm formed in the Mid-Atlantic the possible path and intensity of the storm were modeled and predicted in a range of possible paths. Weather forecasting used a number of real time sources and modeling. My response was in response to Dave Hayword's foolish statement, '"Earlier on today, apparently, a woman rang the BBC and said she heard there was a hurricane on the way... well, if you're watching, don't worry, there isn't!". That evening, the worst storm to hit South East England for three centuries caused record damage and killed 19 people." Real time monitoring and computer modeling were used to follow the track of the storm through is whole path.

When we are talking modeling we mean being able to model the weather for the planet, and accurately predict future storms and weather patterns well before they occur. We can't do that. We probably never will be able to do that. Small initial variables can result in wildly different outcomes. And there is no way to know what all of those initial variables are.

No, that is not the main purpose of computer modeling in weather prediction. Modeling for weather forecasting is short term, like three to ten days in advance. Read the article. You're contorting this in a confusing way. There is a difference between modeling for weather forecasting, long term weather prediction ( six months to a year), and modeling climate trends for long term predictions.

[shunyadragon]

Computer modeling in hurricane forecasting.

Hurricane Forecast Computer Models

By Dr. Jeff Masters, Director of Meteorology

The behavior of the atmosphere is governed by physical laws which can be expressed as mathematical equations. These equations represent how atmospheric quantities such as temperature, wind speed and direction, humidity, etc., will change from their initial current values (at the present time). If we can solve these equations, we will have a forecast. We can do this by sub-dividing the atmosphere into a 3-D grid of points and solving these equations at each point. These models have three main sources of error:

1) Initialization: We have an imperfect description of what the atmosphere is doing right now, due to lack of data (particularly over the oceans). When the model starts, is has an incorrect picture of the initial state of the atmosphere, so will always generate a forecast that is imperfect.

2) Resolution: Models are run on 3-D grids that cover the entire globe. Each grid point represents of piece of atmosphere perhaps 40 km on a side. Thus, processes smaller than that (such as thunderstorms) are not handled well, and must be "parameterized". This means we make up parameters (fudge factors) that do a good job giving the right forecast most of the time. Obviously, the fudge factors aren't going to work for all situations.

3) Basic understanding: Our basic understanding of the physics governing the atmosphere is imperfect, so the equations we're using aren't quite right.

[shunyadragon]

The Weather & Chaos project of the University of Maryland, College Park (UMCP) was started by Distinguished Professor Eugenia Kalnay of Department of Atmospheric and Oceanic Science and Distinguished Professor Jim Yorke of the Institute for Phyical Science and Technology in 2000. They assembled a group of experts to carry out interdisciplinary research in numerical weather prediction, dynamical system theory and scientific computation. The goal was to conduct research that would lead to a better understanding of the behavior of high-dimensional spatio-temporally chaotic systems, such as the atmosphere, and to improved techniques for numerical weather prediction. The initial funding for the project was generously provided by the W. M. Keck Foundation.

Since it was launched in 2000, the project has grown into a major informal research and educational program at UMCP. The project has led to the publication of many papers in leading atmospheric science and physics journals and to over a dozen Ph. D. dissertations in the Atmospheric and Oceanic Science, Physiscs, Mathematics, Applied Mathematics and Scientific Computation and Electrical Engineering doctoral programs.

If you follow the history of weather forecasting, fairly accurate forecasts have gone from 1 to 3 days. Accuracy drops off, but the 7 day forecasts are helpful, but not real accurate. Please differentiate forecasting with long term weather and clime prediction models.

This thread got far from the dialogue concerning AI. One case in point the role of fractal behavior of neuron networks has nothing to do with predictive modeling using fractal modeling. In the brain fractal behavior is how the neuron network functions.

The articles JohnnyP cited are right on topic.

[Sparko]

False once the storm formed in the Mid-Atlantic the possible path and intensity of the storm were modeled and predicted in a range of possible paths. Weather forecasting used a number of real time sources and modeling. My response was in response to Dave Hayword's foolish statement, '"Earlier on today, apparently, a woman rang the BBC and said she heard there was a hurricane on the way... well, if you're watching, don't worry, there isn't!". That evening, the worst storm to hit South East England for three centuries caused record damage and killed 19 people." Real time monitoring and computer modeling were used to follow the track of the storm through is whole path.




No, that is not the main purpose of computer modeling in weather prediction. Modeling for weather forecasting is short term, like three to ten days in advance. Read the article. You're contorting this in a confusing way. There is a difference between modeling for weather forecasting, long term weather prediction ( six months to a year), and modeling climate trends for long term predictions.

what you are describing is predicting POSSIBLE storm paths using computers, not modeling the weather as I said originallly. I made the comment that it was unlikely that we could create a sentient model of the human brain when we can't even model the weather of this planet.

Now you are talking about only assistance in predicting storms using computers. That is to a sentient artificial mind like claiming SIRI proves AI sentience is possible. Nobody disputes that computer modeling can help in many situations or that AI systems are impossible. We are talking SENTIENCE, and self-awareness levels of AI.

[Sparko]

Computer modeling in hurricane forecasting.


3) Basic understanding: Our basic understanding of the physics governing the atmosphere is imperfect, so the equations we're using aren't quite right.

This is exactly my point with AI sentience. we can't even understand the physics of our own atmosphere, so there is no way we are even close to understanding how our brains actually work and create conscious awareness.

[shunyadragon]

This is exactly my point with AI sentience. we can't even understand the physics of our own atmosphere, so there is no way we are even close to understanding how our brains actually work and create conscious awareness.

Let's start a thread on weather and climate and get back to the subject at hand.

[shunyadragon]

what you are describing is predicting POSSIBLE storm paths using computers, not modeling the weather as I said originallly. I made the comment that it was unlikely that we could create a sentient model of the human brain when we can't even model the weather of this planet.

Short term predicting of storm paths is modeling weather for a three to 7 day forecasting as demonstrated in the English hurricane article.

Now you are talking about only assistance in predicting storms using computers. That is to a sentient artificial mind like claiming SIRI proves AI sentience is possible. Nobody disputes that computer modeling can help in many situations or that AI systems are impossible. We are talking SENTIENCE, and self-awareness levels of AI.

These two different subjects, please let's get back to the subject at hand. Weather forecasting, long range forecasting, and climate predictive models are a different subject and application of Chaos and fractal natural behavior then Chaos behavior in neuron networks natural and artificial. In terms of brains and artificial intelligence we are not dealing with forecasting and predictions.

[David Hayward]

Short term predicting of storm paths is modeling weather for a three to 7 day forecasting as demonstrated in the English hurricane article.

I definitely agree with you that the weather can be, and routinely is, modelled, and that the models produce acceptably accurate short-term predictions (which become less accurate as short-term merges into long-term.)

Yes, it was the forecast presenter who presented so strikingly wrongly, not a wrong forecast as such.

These <are> two different subjects, please let's get back to the subject at hand. Weather forecasting, long range forecasting, and climate predictive models are a different subject and application of Chaos and fractal natural behavior than Chaos behavior in neuron networks natural and artificial. In terms of brains and artificial intelligence we are not dealing with forecasting and predictions.

Although weather is chaotic, I question whether it is fractal, and also question whether any of the forecasting models used -- thanks for the interesting link -- use fractals. So I can agree that weather forecasting is irrelevant to the OP, and to discussions of the brain, and to AI, so that it is inappropriate to dwell further on weather forecasting .

[Carrikature]

This is exactly my point with AI sentience. we can't even understand the physics of our own atmosphere, so there is no way we are even close to understanding how our brains actually work and create conscious awareness.

The two are entirely unrelated. It's not a matter of understanding the physics, it's a matter of understanding the complexities of the interaction. With brains, we at least have the luxury of turning on and off different switches to see what happens. Not so with the weather.

[Sparko]

Let's start a thread on weather and climate and get back to the subject at hand.

You are the one who went off track. I was using our lack of knowledge to even model weather as a comparison to how little we know about modeling a brain. Of course, as usual, you go off on the tangent to avoid the topic and when you are proven wrong on your tangent, you want to ignore that and "get back to the subject at hand"

You are a funny guy Frank. You definitely suffer from DK syndrome and are pretty much nothing but a pompous jackass.

[Sparko]

The two are entirely unrelated. It's not a matter of understanding the physics, it's a matter of understanding the complexities of the interaction. With brains, we at least have the luxury of turning on and off different switches to see what happens. Not so with the weather.

Oh you can turn off and on individual neurons to "see what happens" in a brain? If that were even possible, that doesn't really help when you have over 86 billion neurons just in the typical cerebral cortex of a human brain. One individual neuron doesn't tell you much. We lose brain cells all the time.

I don't know if we can ever create a sentient artificial intelligence, but I can pretty much guarantee you that we are nowhere close enough to even understanding how a brain works at this point to even come close at our level of technology. Perhaps someone will have a breakthrough and it will become easy, but barring such, I would think it would be 50 to 100 years before we could come close to creating such an intelligence. If ever. We just don't know yet.

To claim otherwise is just guessing. It is like back at the turn of the last century when they thought they could make "mechanical brains" out of gears and such because they had mechanical calculating machines. In fact it is impossible to build a mechanical intelligence as we know now. It took vacuum tubes and eventually microprocessors to even create basic software programs and expert systems and we are still nowhere close to creating sentience.

[David Hayward]

...I don't know if we can ever create a sentient artificial intelligence, but I can pretty much guarantee you that we are nowhere close enough to even understanding how a brain works at this point to even come close at our level of technology. Perhaps someone will have a breakthrough and it will become easy, but barring such, I would think it would be 50 to 100 years before we could come close to creating such an intelligence. If ever. We just don't know yet.

To claim otherwise is just guessing. It is like back at the turn of the last century when they thought they could make "mechanical brains" out of gears and such because they had mechanical calculating machines. In fact it is impossible to build a mechanical intelligence as we know now. It took vacuum tubes and eventually microprocessors to even create basic software programs and expert systems and we are still nowhere close to creating sentience.

The idea of "mechanical brains" was prevalent in the days of machines; then with the development of telephony, the brain became the enchanted wiring loom; after that the brain became the wetware electronic computer, either hard-wired or running software, or both; some now wish to develop the idea of the brain as being a wetware quantum computer: the cynic would say that whatever idea of the brain is current at a given point in time mirrors the latest fashionable technological advance, but will appear laughable in retrospect.

[Sparko]

The idea of "mechanical brains" was prevalent in the days of machines; then with the development of telephony, the brain became the enchanted wiring loom; after that the brain became the wetware electronic computer, either hard-wired or running software, or both; some now wish to develop the idea of the brain as being a wetware quantum computer: the cynic would say that whatever idea of the brain is current at a given point in time mirrors the latest fashionable technological advance, but will appear laughable in retrospect.

exactly.