Discussion
JustinP1 said:
You miss the point.
The point is that if development was really exponential, and especially at the rate that you've mentioned, then the reality would be development far past where even science fiction writers predicted.
It's not.
Rubbish - we're to take our future timeline predictions from Science Fiction thriller films and writers now are we? The lack of a Time Machine now is your evidence for non-Exponential growth is it The point is that if development was really exponential, and especially at the rate that you've mentioned, then the reality would be development far past where even science fiction writers predicted.
It's not.
Growth also isn't all about the hardware its also about the software. It's also about fields other than Space and Time Travel (I can't believe you referenced Back To The Future
)Take Biology, it took 13 years to decode the first human genome - now they can decode an individuals genome in less than 2 hours.
However if you're taking your cues for the future from Terminator type films which is really quite funny then on that note Drone strikes will shortly be fully automated. Here's something almost straight from Terminator for you and it's already 2 years old.
Christ, you can have your own personal Drone for under £1,000 and they're about to release drones that can follow you about filming you in your car.
Cast your mind back merely 15 years to see where we were then and are now.
Not sure why this subject has raised friction, nobody knows the answer. Yet.
I really enjoy reading about this topic, as I think it is/could be very exciting and I'm happy to hear about it from both sides... not sure it needs to descend into belittling someone elses thoughts though. Guess that's the PH way?!
I really enjoy reading about this topic, as I think it is/could be very exciting and I'm happy to hear about it from both sides... not sure it needs to descend into belittling someone elses thoughts though. Guess that's the PH way?!
JustinP1 said:
Ok Mudflaps, you stock up on your tins for when the robots take over.
I won't, and we'll see who's typing here in ten years time...!?
I'm with Mudflaps on this. I don't think the introduction of AI will take the form of robots which is what popular scifi has pushed over the years (that's not to say we won't have robots). It will come is subtle ways that are already happening. Look at talk to Google or Siri. This is narrow AI that learns from us, the likes of which will increasingly pervade through society without many of us even knowing. We will be talking to PC's with limited use of keyboards in 10 years.I won't, and we'll see who's typing here in ten years time...!?
JustinP1 said:
mudflaps said:
JustinP1 said:
Ok Mudflaps, you stock up on your tins for when the robots take over.
I won't, and we'll see who's typing here in ten years time...!?
Typing?I won't, and we'll see who's typing here in ten years time...!?
How quaint.
RAM?
Processing Power?
Data Storage?
Cloud Computing?
Pixels per Dollar?
Even research into Computing?
And all of that is just in the field of computing and doesn't cover it all.
Biological genome databases are growing exponentially as are many other fields in many other disciplines.
Ironically for further information you can literally 'Ask' Google.
Processing Power?
Data Storage?
Cloud Computing?
Pixels per Dollar?
Even research into Computing?
And all of that is just in the field of computing and doesn't cover it all.
Biological genome databases are growing exponentially as are many other fields in many other disciplines.
Ironically for further information you can literally 'Ask' Google.
Cloud/Big Data/Data Storage are currently fashionable so there's bound to be an increase in research & spending. I'm sure we could find similar graphs for just about any research discipline - and then show how they tail off as the next wave of fashionable research takes over.
Moore's law is slowing - it's no longer a doubling in density every 18 months, it's 2 years or more as we're close to the physical limits of the materials used in chip manufacture.
The biggest barrier, though, is that we still don't really understand what intelligence is. We have definitions of how to judge whether or not it's present and in roughly what quantity, but not real mechanisms for how it works.
You can have all the processing power and storage you like, but if you don't understand what you're trying to do with it, you won't achieve anything.
Moore's law is slowing - it's no longer a doubling in density every 18 months, it's 2 years or more as we're close to the physical limits of the materials used in chip manufacture.
The biggest barrier, though, is that we still don't really understand what intelligence is. We have definitions of how to judge whether or not it's present and in roughly what quantity, but not real mechanisms for how it works.
You can have all the processing power and storage you like, but if you don't understand what you're trying to do with it, you won't achieve anything.
Narrow or weak AI is seen in abundance. Asking Siri directions to the nearest restaurant and getting it to understand you maybe 80% of the time is a relative cake-walk.
Narrow AI doesn't need anything more complex then for SIRI to pattern match certain keywords. SIRI doesn't understand what you are asking it, what a restaurant is or why you'd even want to go there.
STRONG AI requires a lot more than just faster processor chips, more storage and a large database. It requires huge advancement in many areas including a fundamental understanding of how the brain works which we are nowhere close to yet.
Don't get me wrong, I am all for progress and would love to be proven wrong but I very much doubt we will be seeing anything approaching Strong AI within the next 20 years. Kursweil's theory while very interesting discounts one very important thing. The rule of diminishing returns. Yes we have made fantastic progress over the last few hundred years which could lead one to think that we are on an exponential curve, however we have caught all the low hanging fruit up to this point, from now on we are moving into areas where there are no clear cut or easy answers. In fact for the last 10-15 years I'd say we've hit a period of stagnation in terms of new discoveries and not just improvements on stuff we already know. Sure we have had small successes in labs in various fields but turning even those small discoveries into viable real world applications is predicted to take decades.
PS Google deciding to call something Quantum doesn't make it so, Quantum computing isn't even in it's infancy, they've just about managed to write 1 quantum bit in a lab using huge magnets, they'll need to write millions to make a quantum computer viable and that's only one of the many hundreds of challenges quantum computing faces. We don't even fully understand how quantum particles work yet!
Narrow AI doesn't need anything more complex then for SIRI to pattern match certain keywords. SIRI doesn't understand what you are asking it, what a restaurant is or why you'd even want to go there.
STRONG AI requires a lot more than just faster processor chips, more storage and a large database. It requires huge advancement in many areas including a fundamental understanding of how the brain works which we are nowhere close to yet.
Don't get me wrong, I am all for progress and would love to be proven wrong but I very much doubt we will be seeing anything approaching Strong AI within the next 20 years. Kursweil's theory while very interesting discounts one very important thing. The rule of diminishing returns. Yes we have made fantastic progress over the last few hundred years which could lead one to think that we are on an exponential curve, however we have caught all the low hanging fruit up to this point, from now on we are moving into areas where there are no clear cut or easy answers. In fact for the last 10-15 years I'd say we've hit a period of stagnation in terms of new discoveries and not just improvements on stuff we already know. Sure we have had small successes in labs in various fields but turning even those small discoveries into viable real world applications is predicted to take decades.
PS Google deciding to call something Quantum doesn't make it so, Quantum computing isn't even in it's infancy, they've just about managed to write 1 quantum bit in a lab using huge magnets, they'll need to write millions to make a quantum computer viable and that's only one of the many hundreds of challenges quantum computing faces. We don't even fully understand how quantum particles work yet!
mudflaps said:
RAM?
Processing Power?
Data Storage?
Cloud Computing?
Pixels per Dollar?
Even research into Computing?
And all of that is just in the field of computing and doesn't cover it all.
Biological genome databases are growing exponentially as are many other fields in many other disciplines.
Ironically for further information you can literally 'Ask' Google.
It's been a good few year since A-Level Maths but, none of the plotted graphs show an exponential curve...?Processing Power?
Data Storage?
Cloud Computing?
Pixels per Dollar?
Even research into Computing?
And all of that is just in the field of computing and doesn't cover it all.
Biological genome databases are growing exponentially as are many other fields in many other disciplines.
Ironically for further information you can literally 'Ask' Google.
They are all linear, save for an odd outlier which is included where an unbiased analyst would ignore.
The one about how computing will match the human brain is particularly skewed, and despite there not being a point plotted for the last 15 years, sees exponential growth where there is no evidential proof for this.
Edited by JustinP1 on Friday 10th July 13:05
marshalla said:
Moore's law is slowing - it's no longer a doubling in density every 18 months, it's 2 years or more as we're close to the physical limits of the materials used in chip manufacture.
It has periods of slower growth and then faster growth, the trend is all. Yes, a silicon based chips physical limits are now on the horizon but everybody thought processors with components 20nm wide would be the limit, then 14nm wide but IBM have just announced 7nm has now been achieved in their lab.
Silicon chips are also only the 5th paradigm in computing and as we reach the limits of one paradigm another has always arisen out of necessity. There are alternatives to Silicon being developed as I type this.
marshalla said:
The biggest barrier, though, is that we still don't really understand what intelligence is. We have definitions of how to judge whether or not it's present and in roughly what quantity, but not real mechanisms for how it works.
I'm not sure that I agree that we are the best judges of what intelligence is or isn't hence the warnings now coming out from scientists and scholars alike.Anyway this thread is really off topic now so perhaps this subject is for a more dedicated thread.
JustinP1 said:
It's been a good few year since A-Level Maths but, none of the plotted graphs show an exponential curve...?
They are all linear, save for an odd outlier which is included where an unbiased analyst would ignore.
Wrong.They are all linear, save for an odd outlier which is included where an unbiased analyst would ignore.
Look at the Y axis for (say) the first graph. If that doesn't explain it then I can't help you.
JustinP1 said:
mudflaps said:
JustinP1 said:
It's been a good few year since A-Level Maths but, none of the plotted graphs show an exponential curve...?
They are all linear, save for an odd outlier which is included where an unbiased analyst would ignore.
Wrong.They are all linear, save for an odd outlier which is included where an unbiased analyst would ignore.
Look at the Y axis for (say) the first graph. If that doesn't explain it then I can't help you.
Are you suggesting that's an exponential curve?
Here is a graph showing exponential growth
Here is another but because the Y axis would be sooooooo tall it's had to be truncated as shown in the scaling
Do you see how the Y axis scale is no longer linear in increment ie 200, 400, 600, 800, 1000 etc but is now 10, 100, 1000, 10000 etc?
Now imagine if the RAM graph was using the same scaling (200, 400, 600 etc) for its Y axis, what would the curve look like? The scale has been truncated and the curve is a diagonal line because the data wouldn't fit on a linear scaled graph and, more importantly, the author of the graph expects you to understand this by looking at the data.
In short, you are completely mis-reading all of the graphs. The fact that these graphs ALL use the word 'Exponential' ought to give you a clue.
Guvernator said:
Pistom said:
All I want to know is when can I buy a shagable synth like on the telly.
Not the type that's going to kill me though.
Some reckon it will be in the next 20 years, I'd suggest it will be more like 50+ at least so not in my lifetime unfortunately. Not the type that's going to kill me though.
Yes, I'm speaking for myself here.
Gassing Station | TV, Film, Video Streaming & Radio | Top of Page | What's New | My Stuff