Codops, Singularitarians, Mind uploading, Science, and Science Fiction (Part 4)

Businesses and Watson, Artificial Brains in the work environment, Artificial Brains at home

So, I do congratulate you, reader, for coming this far or joining in the story at this point.

We have covered the idea of the singularity and hopefully some of the mechanics of how the singularity will happen.

Today, there is data science.  A field that I would love to break in to and make my profession or have it be tightly integrated in to my work.  There is already machine learning out there and there are many regulations that businesses have to navigate on a daily basis.

Businesses lose money often when their staff or lawyers make mistakes, say in the application of Sarbanes-Oxley or any of the other laws that impact businesses.

Watson would be a key tool for businesses to have and ask it questions.  This version of Watson would be well versed in all commercial laws, common business practices, and the data from a company’s history.  Business executives would ask it questions instead of whole departments of analysts and based on Watson’s answers action plans would be put in to place.

The impact of a Watson on business should not be underestimated.

When is it, though, that Watson level computing will be available to businesses?

I have been a project manager at a large hospital system and managed the upgrade of SAP from ECC 6.0 to ECC 6.0 with EHP (Enhancement Pack) 5.  This involved the replacement of the entire hardware infrastructure that SAP was installed.  Now, the company I worked at went the highest quality, highest cost, and highest supported hardware so the hardware cost was expensive – even to the point of excessive.

The dollar amount was $500,000 for the upgrade of the hardware to support the SAP Enhancement Pack upgrade.  This included things like $27,000 for 2TB of SAN storage.  Incredibly exorbitant pricing considering I just saw a 2TB portable hard drive (portables typically more expensive than fixed desktop hard drives) for $85 just the other day.  I could literally afford 317 $85 2TB hard drives for the cost of 2TB of SAN hard disk space.  I am sure I am in the wrong business.

In any case, disparaging the costs of hardware in the business environment is not why we are here today.

The question is: based on an initial cost of $500,000 for upgrading hardware, when will a approximately 4 billion dollar company with around 20,000 employees be able to afford Watson level computing.

We know from my previous post that a large family that has computers, mobile devices, and tablets for most members would be able to afford a Watson level computer at around 2036 or 2037.

SAP_To_WatsonGraph 3

You can see on the graph above that Watson’s declining costs intersect the increasing costs of an SAP upgrade with replacement of hardware right on top of the year 2020.  Then Watson’s cost keep declining and the budgeting folks keep increasing with inflation the amount of money to be spent on systems like SAP.  Watson becomes a commodity to businesses 5 years after the first Watson joins the large business arsenal of computing capabilities.

The direct implication is that rapidly after having 1 Watson (assuming it proves itself capable) there will be many Watsons in the work place – learning and then performing different jobs and answering everyone’s questions about the business, the laws relating to the business, and the next steps for that business.

There are, of course, implications in the work place.  We aren’t talking ‘fries up’ fast food restaurant workers, but the major part of the middle class – business analysts, mid-to-high management, and knowledge workers losing their jobs to thousands to millions of Watsons.

It will start happening just 5 years from now.  Businesses will operate better, more efficiently, and more profitably than ever before with fewer major workforce costs.  I can’t say I mourn all the incessant wastes of time that are meetings, but meetings will for the most part be obsolete in terms of dissemination of knowledge, decision making, and forming a consensus or at least an agreement about a specific task, function, or decision.  The workers that attend these meetings cost a lot more than the low level workers and I highly doubt that once one company somewhere has shifted to a legion of Watsons and saved huge sums of money and increased their bottom line – that many companies will resist the urge to have their own legion of Watsons.

Not to mention that I’m not sure what will replace SAP, but with a legion of Watsons on my side, it will most assuredly be replaced by something better, less costly, and more transparent.  There is also a heavy implication that software like SAP, Oracle Apps, and major ERP systems – will all vastly change in only 5 to 10 years.

Finally, we arrive at the artificial brain.  An AI intelligence that is a copy of a human being.  One of the core concepts of this blog – the computerized doppelganger or codop.

Based on Ray Kurzweil’s estimate, the codop should be possible around 2030.  Just to lay in some ground work – I’m going to say that the hardware cost in the future will be similar to the hardware costs of Watson, only increased for inflation.  It is admittedly a position that may not be accurate; however, it is a start.  A place where we can say – if everything stays similar, then we can make some prognostications with some level of confidence.

If we say that Watson’s hardware cost 3 million dollars in 2011 (as it has been indicated on various sites) – then increasing for inflation from 2011 to 2030 by 3.5% per year we arrive at an initial hardware cost of $5,767,504.  Yes, of course, for development of how the brain structures operate, how they function, and a whole host of things there will be additional labor costs.  That isn’t what we are talking about here though.  Those costs will be covered by some entrepreneuring company – and they will charge license fees for the use of that knowledge.  After the codop system has been developed, primarily costs reside with the hardware and the hardware’s drop in cost over time.

The population of the world will be many billions of humans, 1 codop, and 1 inhu (individual human – the person a codop is based).  As great as Watson is; this first codop will likely be the lead scientist in the research for the creation of codops – doubling that person’s ability to do research and gain knowledge on how codops should work.  That happens at 2030 – that is our premise.

SAP_Watson_Codop

Graph 4.

In looking at Graph 4 you may feel a little twinge.  As home computing power increases to Watson levels in 2036, businesses start to get codop capability.

Now, the unemployment problems get real.

Suppose as a business there are tasks that are not suited to the many (hundreds?, thousands?) of Watsons you have crunching data, answering questions, and helping the company.  Perhaps you write software code and it just can’t be done by Watsons.  Perhaps your company is a pharmaceutical company and while Watsons can answer questions based on existing data – they turn out to not be able to learn new things on the basis of that data in science.  Pharmaceutical companies, materials companies, chemistry companies, and any number of science based organizations that need to perform experiments to create new technologies to make a better world.  What do you do?  The rest of the businesses in the world have taken huge advantage of the Watsons and profits are soaring – but not yours.

You copy the people that can make those discoveries, conduct those experiments, and make leaps of logic that are not necessarily inherent in the knowledge base and sometimes turn out to be right.

You make codops of these people so that their work flow can be increased at rates that are unheard of before.  Even business analysts might benefit.  Some companies may not want to use Watsons, but a codop of your business analyst or data scientists is nearly the same as the original person.  There could be an implicit trust between the upper management and the ‘person’ doing the work.

Companies will finance the second wave of codops.  The first wave of codops will be the scientists creating the codop technology and other scientists that the codop creator deems necessary for the survival of humanity.  There will be some sales of codop throughput early on to the very rich to finance the creation of scientist codops for the furthering of human kind.

Year 1 of the codops, and there will only be 1 codop.  Year 2 there will be two codops.  Year 3 there will be 4, year 4 there will be 8, year 5 there will be 16….  Anyone familiar with computer technology knows this pattern of events.

(2040) 10 years after the first codop is born, there will be 1024 codops.

(2050) 20 years after the first codop is born, there will be 1,048,576 codops.  By this time not all codops will be scientists.

(2060) 30 years after the first codop is born, there will be 1,073,741,824 (over a billion).

(2063) 33 years after the first codop is born, there will be more codops functioning than humans alive – 8,589,934,592 – or close to it.

Codops of the best in their fields will be created – probably hundreds of times – making for the final push of many technical fields to loss of employment.

Writers will be able to make copies – and ensure that no SF series is ever left incomplete.

PersonalComputingcodop

 

We can see on this graph the earlier data – Watson level computing in homes by 2036/2037 – and even more, codop level computing in homes around 2055 – 2057.  Dad has a lot of work to do today; however, he and his codop finished it in 2 hours flat and dad gets to play with his kids.

There will be a lot of benefit to personal codops.  The codops can quickly handle household finance and the inhu can simply go on with life.

There are a lot of very intelligent people out there that have grave warnings about AI.  The first generation of AI – will be us – inhu that have been created in to codops.  If we make laws – potentially starting now – in regard to personhood of codops, human/inhuman rights, laws, inheritance, illegal to murder codops, etc – we have a chance to treat codops the way we ourselves want to be treated.

Ahhhh, but with codops finally, we have a chance to do things that are not possible for inhu or humans – multi-thousand year journeys throughout the universe – because when your brain is hardware and software – you can live forever – and perhaps return home to tell the rest of humanity what was out there. 😉

 

 

Below, data for Graph 3, which is really just an appended column to the previous dataset.  This is an inflation increased cost of SAP hardware for a major upgrade by 3.5% per year – and a starting cost of $500,000 based on my experience managing such an SAP upgrade.

Year Business SAP Upgrade Hardware Cost
2011
2012
2013 500000
2014 517500
2015 535613
2016 554359
2017 573762
2018 593843
2019 614628
2020 636140
2021 658405
2022 681449
2023 705299
2024 729985
2025 755534
2026 781978
2027 809347
2028 837674
2029 866993
2030 897338
2031 928745
2032 961251
2033 994894
2034 1029716
2035 1065756
2036 1103057
2037 1141664
2038 1181622
2039 1222979
2040 1265784
2041 1310086
2042 1355939
2043 1403397
2044 1452516
2045 1503354
2046 1555971
2047 1610430
2048 1666795
2049 1725133
2050 1785513
2051 1848006
2052 1912686
2053 1979630
2054 2048917
2055 2120629
2056 2194851
2057 2271671
2058 2351179
2059 2433471
2060 2518642
2061 2606794
2062 2698032
2063 2792463
2064 2890200
2065 2991357
2066 3096054
2067 3204416
2068 3316571
2069 3432651
2070 3552793
2071 3677141
2072 3805841
2073 3939045
2074 4076912
2075 4219604
2076 4367290
2077 4520145
2078 4678350
2079 4842093
2080 5011566

Data for Graph 4 (added to original data columns)

Year Cost
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030 5767504
2031 4844703
2032 4069551
2033 3418423
2034 2871475
2035 2412039
2036 2026113
2037 1701935
2038 1429625
2039 1200885
2040 1008744
2041 847345
2042 711769
2043 597886
2044 502225
2045 421869
2046 354370
2047 297670
2048 250043
2049 210036
2050 176430
2051 148202
2052 124489
2053 104571
2054 87839.7
2055 73785.3
2056 61979.7
2057 52062.9
2058 43732.9
2059 36735.6
2060 30857.9
2061 25920.6
2062 21773.3
2063 18289.6
2064 15363.3
2065 12905.1
2066 10840.3
2067 9105.87
2068 7648.93
2069 6425.1
2070 5397.09
2071 4533.55
2072 3808.18
2073 3198.87
2074 2687.05
2075 2257.13
2076 1895.99
2077 1592.63
2078 1337.81
2079 1123.76
2080 943.957

Code in R for all graphs using the data embedded in pages (in a single spreadsheet and sheet as a csv file):

FutureHomeComputingCosts <- function() {
 PC <- read.csv("C:/Users/Alexander/Dropbox/Creations/Writing/Future/Costs_AB_Watson_Computers_20150424.csv")
 
 plot(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, type="n", ylab="Cost", xlab="Year", yaxt="n")
 
 aty <- seq(par("yaxp")[1], par("yaxp")[2], (par("yaxp")[2] - par("yaxp")[1])/par("yaxp")[3])
 axis(2, at=aty, labels=format(aty, scientific=FALSE), hadj=0.9, cex.axis=0.7, las=2)
 axis(4, c(10000, 35000,100000,150000,170000), cex.axis=0.6, hadj=0.5, las=3)
 
 lines(PC$Year, PC$HC.Low, col="red")
 lines(PC$Year, PC$HC.Medium, col="blue")
 lines(PC$Year, PC$HC.High ,col="green")
 lines(PC$Year, PC$HC.Gamer, col="yellow")
 lines(PC$Year, PC$Author.Home.Computers, col="orange")
 lines(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, col="purple")
 
 costTypes <- c("Low Cost", "Medium Cost", "High Cost", "Gamer Cost", "Family of 6 Computers", 
 "Family of 6 Computers and Mobile Devices")
 costColor <- c("red", "blue", "green", "yellow", "orange", "purple")
 
 legend("topleft", costTypes, lty=c(1,1), col=costColor)
 mtext("Cost of Home Computing Power", side = 3)
 
}
FutureHomeComputingCostsPlusWatson <- function() {
 PC <- read.csv("C:/Users/Alexander/Dropbox/Creations/Writing/Future/Costs_AB_Watson_Computers_20150424.csv")
 
 plot(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, type="n", ylab="Cost", xlab="Year", yaxt="n", xaxt="n")
 
 aty <- seq(par("yaxp")[1], par("yaxp")[2], (par("yaxp")[2] - par("yaxp")[1])/par("yaxp")[3])
 axis(2, at=aty, labels=format(aty, scientific=FALSE), hadj=0.9, cex.axis=0.7, las=2)
 axis(4, c(10000, 35000,100000,150000,170000), cex.axis=0.6, hadj=0.5, las=3)
 axis(1, c(2010, 2015, 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055, 2060, 2065, 2070, 2075, 2080), cex.axis=0.75)

 costTypes <- c("Low Cost", "Medium Cost", "High Cost", "Gamer Cost", "Family of 6 Comps.", 
 "Family of 6 Comps and Devices", "Watson")
 costColor <- c("red", "blue", "green", "yellow", "orange", "purple", "red")
 
 legend("left", costTypes, lty=c(1,1), col=costColor)
 
 lines(PC$Year, PC$HC.Low, col="red")
 lines(PC$Year, PC$HC.Medium, col="blue")
 lines(PC$Year, PC$HC.High ,col="green")
 lines(PC$Year, PC$HC.Gamer, col="yellow")
 lines(PC$Year, PC$Author.Home.Computers, col="orange")
 lines(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, col="purple")
 lines(PC$Year, PC$Watson.Cost, col="red", type="b")
 

 mtext("Cost of Home Computing Power Increases with Inflation and Decreasing Costs of Watson", side = 3)
 
}
FutureBusinessComputingWatson <- function() {
 PC <- read.csv("C:/Users/Alexander/Dropbox/Creations/Writing/Future/Costs_AB_Watson_Computers_20150424.csv")
 
 plot(PC$Year, PC$Business.SAP.Upgrade.Hardware.Cost, type="n", ylab="Cost", xlab="Year", yaxt="n", xaxt="n")
 
 aty <- seq(par("yaxp")[1], par("yaxp")[2], (par("yaxp")[2] - par("yaxp")[1])/par("yaxp")[3])
 axis(2, at=aty, labels=format(aty, scientific=FALSE), hadj=0.9, cex.axis=0.7, las=2)
 axis(1, c(2010, 2015, 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055, 2060, 2065, 2070, 2075, 2080), cex.axis=0.75)
 
 costTypes <- c("SAP", "Watson")
 costColor <- c("black", "red")
 
 legend("topleft", costTypes, lty=c(1,1), col=costColor)
 
 lines(PC$Year, PC$Business.SAP.Upgrade.Hardware.Cost, col="black")
 lines(PC$Year, PC$Watson.Cost, col="red", type="b")
 
 
 mtext("Cost of Business Computing Power Increases with Inflation and Decreasing Costs of Watson", side = 3)
 
}
FutureBusinessComputingCodop <- function() {
 PC <- read.csv("C:/Users/Alexander/Dropbox/Creations/Writing/Future/Costs_AB_Watson_Computers_20150424.csv")
 
 plot(PC$Year, PC$Business.SAP.Upgrade.Hardware.Cost, type="n", ylab="Cost", xlab="Year", yaxt="n", xaxt="n")
 
 aty <- seq(par("yaxp")[1], par("yaxp")[2], (par("yaxp")[2] - par("yaxp")[1])/par("yaxp")[3])
 axis(2, at=aty, labels=format(aty, scientific=FALSE), hadj=0.9, cex.axis=0.7, las=2)
 axis(1, c(2010, 2015, 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055, 2060, 2065, 2070, 2075, 2080), cex.axis=0.75)
 
 costTypes <- c("SAP", "Watson", "codop")
 costColor <- c("black", "red", "green")
 
 legend("topleft", costTypes, lty=c(1,1), col=costColor)
 
 lines(PC$Year, PC$Business.SAP.Upgrade.Hardware.Cost, col="black")
 lines(PC$Year, PC$Watson.Cost, col="red", type="b")
 lines(PC$Year, PC$Cost, col="green", type="b")
 
 
 mtext("SAP increasing hardware cost vs Decreasing Costs of Watson and the Artificial Brain", side = 3)
 
}
FuturePersonalComputingCodop <- function() {
 PC <- read.csv("C:/Users/Alexander/Dropbox/Creations/Writing/Future/Costs_AB_Watson_Computers_20150424.csv")
 
 PC <- subset(PC, PC$Year > 2020 & PC$Year < 2071)
 
 plot(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, type="n", ylab="Cost", xlab="Year", yaxt="n", xaxt="n")
 
 aty <- seq(par("yaxp")[1], par("yaxp")[2], (par("yaxp")[2] - par("yaxp")[1])/par("yaxp")[3])
 axis(2, at=aty, labels=format(aty, scientific=FALSE), hadj=0.9, cex.axis=0.7, las=2)
 axis(4, c(35000, 70000, 100000), cex.axis=0.6, hadj=0.5, las=3)
 axis(1, c(2015, 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055, 2060, 2065, 2070, 2075), cex.axis=0.75)
 
 costTypes <- c("Low Cost", "Medium Cost", "High Cost", "Gamer Cost", "Family of 6 Comps.", 
 "Family of 6 Comps and Devices", "Watson", "codop")
 costColor <- c("red", "blue", "green", "yellow", "orange", "purple", "red", "green")
 
 legend("topleft", costTypes, lty=c(1,1,1,1,1,1,4,4), col=costColor)
 
 lines(PC$Year, PC$HC.Low, col="red")
 lines(PC$Year, PC$HC.Medium, col="blue")
 lines(PC$Year, PC$HC.High ,col="green")
 lines(PC$Year, PC$HC.Gamer, col="yellow")
 lines(PC$Year, PC$Author.Home.Computers, col="orange")
 lines(PC$Year, PC$Author.Home.Computers.and.Mobile.Devices, col="purple")
 lines(PC$Year, PC$Watson.Cost, col="red", type="b")
 lines(PC$Year, PC$Cost, col="green", type="b")
 
 
 mtext("Personal computing, Watson, and the codop/artificial brain", side = 3)
 
}