The world of automation
#1156
Re: The world of automation
Gary, Beoz, interested to hear your views on this if you have time to watch...
I've just finished his second book (Homo Deus) and this Davos talk puts some of his concerns about the evolution of data (and humans!) into clearer perspective. Quite a dystopian future is he's right.
I've just finished his second book (Homo Deus) and this Davos talk puts some of his concerns about the evolution of data (and humans!) into clearer perspective. Quite a dystopian future is he's right.
#1157
Re: The world of automation
He gets quite few things wrong IMHO. For a start, data isn't the most important/valuable thing. It's actually at the bottom of a pyramid that goes up to wisdom. The whole 'hacking a human' kind of misses the point, and is built on quite a lot of handwaving I don't think he really understands. For a start, 'organisms are damn well not algorithms' - not even close. You wouldn't even get away with saying they were complex adaptive systems, though you'd be closer. They are not deterministic.
Oh and the 'centralised processing for a better soviet' is from somepoint in the 1980s.
Most of it seems to get drawn from the work of others, particularly this (note the date)
https://edoras.sdsu.edu/~vinge/misc/singularity.html
It's putting a pretty face on old ideas.
Historians just aren't very good at this sort of thing, they seem to think you can bullsh*t in science etc. as you can in history, and get away with it if it 'sounds good'.
There are some things you can do along these lines - just talk to the marketeers. However it's short term and situational - at the individual or group level.
It shades into Asimov's Foundation series - but here's a small practical example. Ireland is currently in the process of shifting from a catholic dominated theocracy, to a more secular western democracy. The drivers for that change have been around for a while, but exactly WHEN then change happens, and how fast, are complex, chaotic, events that you could have all the data in the world and you couldn't predict.
Oh and the 'centralised processing for a better soviet' is from somepoint in the 1980s.
Most of it seems to get drawn from the work of others, particularly this (note the date)
https://edoras.sdsu.edu/~vinge/misc/singularity.html
It's putting a pretty face on old ideas.
Historians just aren't very good at this sort of thing, they seem to think you can bullsh*t in science etc. as you can in history, and get away with it if it 'sounds good'.
There are some things you can do along these lines - just talk to the marketeers. However it's short term and situational - at the individual or group level.
It shades into Asimov's Foundation series - but here's a small practical example. Ireland is currently in the process of shifting from a catholic dominated theocracy, to a more secular western democracy. The drivers for that change have been around for a while, but exactly WHEN then change happens, and how fast, are complex, chaotic, events that you could have all the data in the world and you couldn't predict.
#1158
Lost in BE Cyberspace
Joined: Dec 2010
Posts: 14,040
Re: The world of automation
Gary, Beoz, interested to hear your views on this if you have time to watch...
I've just finished his second book (Homo Deus) and this Davos talk puts some of his concerns about the evolution of data (and humans!) into clearer perspective. Quite a dystopian future is he's right.
https://www.youtube.com/watch?v=hL9uk4hKyg4
I've just finished his second book (Homo Deus) and this Davos talk puts some of his concerns about the evolution of data (and humans!) into clearer perspective. Quite a dystopian future is he's right.
https://www.youtube.com/watch?v=hL9uk4hKyg4
#1160
Re: The world of automation
He gets quite few things wrong IMHO. For a start, data isn't the most important/valuable thing. It's actually at the bottom of a pyramid that goes up to wisdom. The whole 'hacking a human' kind of misses the point, and is built on quite a lot of handwaving I don't think he really understands. For a start, 'organisms are damn well not algorithms' - not even close. You wouldn't even get away with saying they were complex adaptive systems, though you'd be closer. They are not deterministic.
Oh and the 'centralised processing for a better soviet' is from somepoint in the 1980s.
Most of it seems to get drawn from the work of others, particularly this (note the date)
https://edoras.sdsu.edu/~vinge/misc/singularity.html
It's putting a pretty face on old ideas.
Historians just aren't very good at this sort of thing, they seem to think you can bullsh*t in science etc. as you can in history, and get away with it if it 'sounds good'.
There are some things you can do along these lines - just talk to the marketeers. However it's short term and situational - at the individual or group level.
It shades into Asimov's Foundation series - but here's a small practical example. Ireland is currently in the process of shifting from a catholic dominated theocracy, to a more secular western democracy. The drivers for that change have been around for a while, but exactly WHEN then change happens, and how fast, are complex, chaotic, events that you could have all the data in the world and you couldn't predict.
Oh and the 'centralised processing for a better soviet' is from somepoint in the 1980s.
Most of it seems to get drawn from the work of others, particularly this (note the date)
https://edoras.sdsu.edu/~vinge/misc/singularity.html
It's putting a pretty face on old ideas.
Historians just aren't very good at this sort of thing, they seem to think you can bullsh*t in science etc. as you can in history, and get away with it if it 'sounds good'.
There are some things you can do along these lines - just talk to the marketeers. However it's short term and situational - at the individual or group level.
It shades into Asimov's Foundation series - but here's a small practical example. Ireland is currently in the process of shifting from a catholic dominated theocracy, to a more secular western democracy. The drivers for that change have been around for a while, but exactly WHEN then change happens, and how fast, are complex, chaotic, events that you could have all the data in the world and you couldn't predict.
I'm also slightly uncomfortable with his insistence that organisms are algorithms, but there might be something in it. Our biology comes down to DNA strings, and increasingly there is this idea that we lack freewill (evidenced by FMRI experiments). Even if algorithm is a clumsy analogy, human psychology does seem to be something that can be modeled with enough data; I mean, we're fundamentally not that different or that unpredictable. In his book I found the idea of being able to know a person's political beliefs as well or better than the person, quite intriguing. It did suggest that the whole idea of elections was anachronistic, and then it quite naturally leads to a "computer rules" scenario.
On the Ireland example, not so sure. I think the shift itself could be predicted, if not the exact date. And what Harari is saying is that as the accumulation of data continues, and (importantly) the processing capability plus the brain science, we may well be able to predict these shifts.
#1161
Re: The world of automation
I'm also slightly uncomfortable with his insistence that organisms are algorithms, but there might be something in it. Our biology comes down to DNA strings, and increasingly there is this idea that we lack freewill (evidenced by FMRI experiments). Even if algorithm is a clumsy analogy, human psychology does seem to be something that can be modeled with enough data; I mean, we're fundamentally not that different or that unpredictable. In his book I found the idea of being able to know a person's political beliefs as well or better than the person, quite intriguing. It did suggest that the whole idea of elections was anachronistic, and then it quite naturally leads to a "computer rules" scenario.
On the Ireland example, not so sure. I think the shift itself could be predicted, if not the exact date. And what Harari is saying is that as the accumulation of data continues, and (importantly) the processing capability plus the brain science, we may well be able to predict these shifts.
On the Ireland example, not so sure. I think the shift itself could be predicted, if not the exact date. And what Harari is saying is that as the accumulation of data continues, and (importantly) the processing capability plus the brain science, we may well be able to predict these shifts.
Put it this way, if you could predict with accuracy and certainty, you could make a killing on the stock market. That's actually a more bounded and predictable scenario than Ireland, and I don't see anyone doing that.
By the way, if you want a better understanding of where things might go, even though its wrong and references are 13 years out of date, try
Accelerando
and then consider that its only 13 years, and just how much is different even today...
#1162
Re: The world of automation
As I said, you can get certain things right, a small distance ahead, or where their is negative feedback keeping it within bounds. But the entire system of your life, or a group, it complex/chaotic. You can only go a small distance into predictions before it gets hopelessly unpredictable.
Put it this way, if you could predict with accuracy and certainty, you could make a killing on the stock market. That's actually a more bounded and predictable scenario than Ireland, and I don't see anyone doing that.
By the way, if you want a better understanding of where things might go, even though its wrong and references are 13 years out of date, try
Accelerando
and then consider that its only 13 years, and just how much is different even today...
Put it this way, if you could predict with accuracy and certainty, you could make a killing on the stock market. That's actually a more bounded and predictable scenario than Ireland, and I don't see anyone doing that.
By the way, if you want a better understanding of where things might go, even though its wrong and references are 13 years out of date, try
Accelerando
and then consider that its only 13 years, and just how much is different even today...
I don't think he's suggesting predictions of life or history, he's suggesting the ability to accurately profile how people will react psychologically and biologically, and building a society on that deep data. Biometric data (not consumer data).
EDIT. Hmmmmm. I can see why Accelerando was published under creative commons...speed read to Chapter 5, and it seems to be just a jargon fest, and not a very good one at that. Hopefully, you read it 13 years ago when you were an impressionable youth.
Last edited by Shard; Feb 1st 2018 at 11:14 am.
#1163
Re: The world of automation
And he freely admits that a technique for writing near future is to hit you with enough new ideas that you buy in to it being it's own world, 5 minutes into the future.
As for it not being a 'very good one', well here's what it won :
https://en.wikipedia.org/wiki/Accele...nd_nominations
Trust me, although there are issues with the model, and certain references have dated, it's actually not too far off the mark of today's reality and where it takes it makes a lot more sense than the historian.
#1164
Re: The world of automation
Well first, it isn't a freebie except for that page, it's a real book from a real (and well known) author.
And he freely admits that a technique for writing near future is to hit you with enough new ideas that you buy in to it being it's own world, 5 minutes into the future.
As for it not being a 'very good one', well here's what it won :
https://en.wikipedia.org/wiki/Accele...nd_nominations
Trust me, although there are issues with the model, and certain references have dated, it's actually not too far off the mark of today's reality and where it takes it makes a lot more sense than the historian.
And he freely admits that a technique for writing near future is to hit you with enough new ideas that you buy in to it being it's own world, 5 minutes into the future.
As for it not being a 'very good one', well here's what it won :
https://en.wikipedia.org/wiki/Accele...nd_nominations
Trust me, although there are issues with the model, and certain references have dated, it's actually not too far off the mark of today's reality and where it takes it makes a lot more sense than the historian.
I'm not totally sold on 'the historian' either, he does make some fairly sweeping assumptions economic systems in order to tie together his vision of data society. But then again, he openly admits it's all conjecture and serves to start the discussion on where we might be going with (even bigger) data. Just this week this Amazon healthcare story points toward a changing model, with further reliance on data and the concomitant privacy/rights issues.
#1165
Re: The world of automation
Okay. Maybe it's just not my cup of tea and/or I don't have the requisite domain knowledge. And didn't ready the whole thing, so it's a bit cheeky to comment.
I'm not totally sold on 'the historian' either, he does make some fairly sweeping assumptions economic systems in order to tie together his vision of data society. But then again, he openly admits it's all conjecture and serves to start the discussion on where we might be going with (even bigger) data. Just this week this Amazon healthcare story points toward a changing model, with further reliance on data and the concomitant privacy/rights issues.
I'm not totally sold on 'the historian' either, he does make some fairly sweeping assumptions economic systems in order to tie together his vision of data society. But then again, he openly admits it's all conjecture and serves to start the discussion on where we might be going with (even bigger) data. Just this week this Amazon healthcare story points toward a changing model, with further reliance on data and the concomitant privacy/rights issues.
So if they have data that you purchased a pregnancy test kit, and then started buying healthier food and not the usual quantity of wine, it can infer what else you might be buying in future and sell at you.
At no point does someone in google know you - they just pattern match and run the probabilities. They don't even really understand how such deep mind structures work. And they don't know that you are suddenly going to be buying figs in maple syrup. All it is is pattern matching by opaque neural nets.
It doesn't create some new species.
Early stage AI research, back in the 60s and 70s, thought they could pull all the data in, put a few algorithms on top, and generate an AI. And it didn't work, except in scenarios where knowledge was already codified and regurgitated (eg doctors) and it failed in the same way that the human doctors did - because it lacked the wisdom that always occurs in the last 10 mins of House.
The automation that's coming over the horizon is just automated pattern matching. It wins because it replaces someone trying to pour over data and turn it into automation algorithms with a machine doing the same - and then just cloning that knowledge cheaply. And that works because we have spent the time since the industrial revolution trying to turn humans into robots who follow defined rules (processes). We took the humanity OUT of the systems of society - and surprise surprise, that means we can take the humans out.
Creativity, innovation, backstabbing corruption, those are still human roles - at least until you point those pattern matching algorithms at what second hand car salesmen say - then you will have the lying b*st*rd neural net you can bolt into other automated processes....
#1166
Re: The world of automation
I think the thing that really rankled me about the historian (besides the science stuff he got wrong whilst blithely stating it as fact), was the insistence that data was the thing. Google don't gather all the data because they want the data. Rather it's that they can point their pattern matching deep learning at it to pull up a neural net that turns the data into knowledge - which they can then use to benefit off of.
So if they have data that you purchased a pregnancy test kit, and then started buying healthier food and not the usual quantity of wine, it can infer what else you might be buying in future and sell at you.
At no point does someone in google know you - they just pattern match and run the probabilities. They don't even really understand how such deep mind structures work. And they don't know that you are suddenly going to be buying figs in maple syrup. All it is is pattern matching by opaque neural nets.
It doesn't create some new species.
Early stage AI research, back in the 60s and 70s, thought they could pull all the data in, put a few algorithms on top, and generate an AI. And it didn't work, except in scenarios where knowledge was already codified and regurgitated (eg doctors) and it failed in the same way that the human doctors did - because it lacked the wisdom that always occurs in the last 10 mins of House.
The automation that's coming over the horizon is just automated pattern matching. It wins because it replaces someone trying to pour over data and turn it into automation algorithms with a machine doing the same - and then just cloning that knowledge cheaply. And that works because we have spent the time since the industrial revolution trying to turn humans into robots who follow defined rules (processes). We took the humanity OUT of the systems of society - and surprise surprise, that means we can take the humans out.
Creativity, innovation, backstabbing corruption, those are still human roles - at least until you point those pattern matching algorithms at what second hand car salesmen say - then you will have the lying b*st*rd neural net you can bolt into other automated processes....
So if they have data that you purchased a pregnancy test kit, and then started buying healthier food and not the usual quantity of wine, it can infer what else you might be buying in future and sell at you.
At no point does someone in google know you - they just pattern match and run the probabilities. They don't even really understand how such deep mind structures work. And they don't know that you are suddenly going to be buying figs in maple syrup. All it is is pattern matching by opaque neural nets.
It doesn't create some new species.
Early stage AI research, back in the 60s and 70s, thought they could pull all the data in, put a few algorithms on top, and generate an AI. And it didn't work, except in scenarios where knowledge was already codified and regurgitated (eg doctors) and it failed in the same way that the human doctors did - because it lacked the wisdom that always occurs in the last 10 mins of House.
The automation that's coming over the horizon is just automated pattern matching. It wins because it replaces someone trying to pour over data and turn it into automation algorithms with a machine doing the same - and then just cloning that knowledge cheaply. And that works because we have spent the time since the industrial revolution trying to turn humans into robots who follow defined rules (processes). We took the humanity OUT of the systems of society - and surprise surprise, that means we can take the humans out.
Creativity, innovation, backstabbing corruption, those are still human roles - at least until you point those pattern matching algorithms at what second hand car salesmen say - then you will have the lying b*st*rd neural net you can bolt into other automated processes....
#1167
Re: The world of automation
I understand that, but he's looking 50-60 years hence. So it's not about marketing, it's about 'total data'. You have to admit, without even invoking an AGI overlord, that once a person's DNA, entire speaking-browsing-writing corpus, geolocation history, social history is known AND analysed against every single other human's 'total data' that some kind of new societal model could emerge? If big tech is left to its own devices, surely this joined up data complete situation will arise. Extrapolating from what's happened in the last few decades. So, the question is how this deep data should be controlled or regulated. In the past (say 15 years ago) the constraint was the ability to analyse and interpret the data, but that constraint is fading fast. His idea is that in the long term machine learning may well lead to the machines knowing us better than ourselves. It seems plausible to me.
Can AI develop or even begin to explain Lifeforce ?
#1170
I still dont believe it..
Joined: Oct 2013
Location: 12 degrees north
Posts: 2,775
Re: The world of automation
I have retired from a lifetime in IT, of all sorts, i even wrote some cobol on paper cards early on in my career!
The last 25 years i have been managing large then huge then vast support pyramids, and all that time i was told that i and my team/dept/group/division/directorate/company would be designed out of existance, that the technological singularity was coming etc etc.
Actually as mankind strived for more complexity, it just got more complex, needing more people and cost more money. Sure things got smarter, they had to, the device you are looking at has an os with 5-8 million lines of code, and the other stuff say 10 times that?
The machines will definitely take over - errm they have already in some ways - witness people lanes for walkers imbedded into their smart phones - telling their soft parts [legs] where to walk [walking gps] but 10 seconds later a bug or virus will see to that!
There are some fascinating sci fi scenarios about what could happen - i commend to you greg bears blood music for example.
Actually i have no fears, i’d love to ask the universal intelligence to make fusion power a reality, i’m just worried that 2 years later it would reply either ‘let there be light’ or ‘the answer is 42’...
The last 25 years i have been managing large then huge then vast support pyramids, and all that time i was told that i and my team/dept/group/division/directorate/company would be designed out of existance, that the technological singularity was coming etc etc.
Actually as mankind strived for more complexity, it just got more complex, needing more people and cost more money. Sure things got smarter, they had to, the device you are looking at has an os with 5-8 million lines of code, and the other stuff say 10 times that?
The machines will definitely take over - errm they have already in some ways - witness people lanes for walkers imbedded into their smart phones - telling their soft parts [legs] where to walk [walking gps] but 10 seconds later a bug or virus will see to that!
There are some fascinating sci fi scenarios about what could happen - i commend to you greg bears blood music for example.
Actually i have no fears, i’d love to ask the universal intelligence to make fusion power a reality, i’m just worried that 2 years later it would reply either ‘let there be light’ or ‘the answer is 42’...