Proving the Singularity is On the Horizon
I will apologize in advance that this is a long post. Hopefully, I can make this in to a YouTube video and it will be short.
How do you prove something that has not happened? Math appears to be a logical course of action. However, that can go awry like the people who were convinced the world was going to end on May 16, 2011. Well, I went skydiving that day and I can tell you as a first-person eye-witness that as I fell to the Earth I did not see anyone ascend. Harold Camping; however, thought that he was backed firmly by numbers and mathematics in predicting the end of the world.
Of course, we should be clear that predicting the singularity is not the same thing as predicting the end of the world. The world you have known for your whole life is always ending moment by moment. It has nothing to do with end of the human species.
Things will change on the day of the singularity, but only faster than they happen today. The day of the singularity on that day will appear no different than the days before or the immediate days afterward; however, when dividing human history in to logical groupings for understanding there could well be a day that we could say, “This is the beginning of the singularity”.
The math used in this will be simple math. The model will have flaws, but it will still enable us to make predictions – even if the specific date might be different than our predicted date – it will definitely be within the ballpark.
We will track the inflationary increases to the costs of computers in present households.
- Household with 1 low cost PC ($500)
- Household with 1 medium cost PC ($1000)
- Household with 1 high cost PC ($2500)
- Household with 1 gamer PC ($5000)
- Author’s household total computing costs ($13,200 – excluding high powered cell phones)
- Author’s household total computing costs including tablets and smart phones. ($15,644 including all mobile devices)
As inflation increases over time of about 3.5% per year – the costs a typical household that purchases these computers would increase. While it is true that PC hardware decreases in costs by about 16% per year, we are going by what the household is willing to pay in today’s dollars and what those today’s dollars will be in the future.
For the purposes of this estimation I am not including smart phones and tablets. This is potentially a source of error as tablet sales increase and traditional PC sales decrease; however, including the costs of tablets and smart phones could potentially throw the numbers in to a broad range of values that may not be representative of anything.
However; in order to include some trace of what the computing costs of tablets and smart phones in to the equation I have made a final group that includes the costs that the author has accumulated with his family of 6 (two adults and 4 children).
The household I live in I would consider to be middle-class. In #5, we include only the computers. This winds up to be $13,200 and including all of our cell phones, tablets, and our computers this goes up to $15,644. This isn’t the costs over the course of 1 year, but the three year investment cost in computing devices.
Here is a graph of the costs of computing for the above levels with additional cost for inflation over the years from 2011 to 2080.
Graph 1 shows the different curves of the increasing prices of computing. As these follow inflation, the end costs are equivalent to today’s prices. This means that if a family of 6 spent $15,644 over the course of two or three years on computers, mobile phones, and tablets in 2011, this would be equivalent to $167,970 in 2080.
Ray Kurzweil uses a point in time when person computing power of $1,000 PC can afford the power to contain a human mind/simulated mind as the point for the singularity. This is a slight misrepresentation as long before home computers can host simulated human minds drastic changes in human society will likely occur.
For example as I mentioned in Part 2, Watson was created and beat the two Jeopardy champions. In essence, a computer of Watson’s capability can answer any question posed to it by any person – potentially in any language. There will be a long time before artificial brains (codops) or any AI system will be available in every household that Watson level computers will be available and owned and perhaps ubiquitous in homes across the world.
Imagine, any question you have, any question your child has in school can be answered by a device in your home. This is far beyond what the internet can do. The internet contains data. The internet can present data. Data requires analysis before it becomes information. Information is required before you can answer questions.
So, let’s add to our graph the cost of Watson’s hardware – declining as previously mentioned – at 16% per year. Yes, it is another rough estimate. More than the specific date when the lines intersect is the idea that these lines will intersect. The follow up idea that long before home computer affordability of Watson level computing comes the affordability of businesses and governments of Watson level computing and problem solving.
You can see from the second graph that the inflation adjusted increasing costs of personal computing intersects with the declining costs (16% per year) of Watson’s hardware costs.
In fact if all of our assumptions are accurate (and there is a lot of room for problems) it appears that sometime in late 2036 or 2037 a middle income family that has 4 kids and lots of computers and mobile devices would be able to afford 1 Watson – if they choose not to buy any other new devices. Based on the data below you can see that in 2036 the inflation increasing costs of the family of 6 and all their computing devices will cost $36,971 compared to the declining cost of Watson costing about: $38,380.
Now it is important to differentiate – that the inflation increasing costs for all the home computers is based on what we are willing to spend today on computers and keeping that constant. The declining costs of Watson is based on the falling costs and increasing capabilities of computer hardware.
It is entirely possible, that the family of 6 might choose to spend less money by 16% per year on computing hardware and more on other products with their disposable income. However, based on people’s habits, the consumption of technology leads to more spending on technology, not less, over time. (I’ll need to support this with some data at a later point in time, so far this is just based on my personal observations).
In addition, computing takes many forms. For example, Siri does not exist on your Apple iPhone. Siri is a form of distributed computing where the consumer/end user does not bear the direct cost of the hardware and infrastructure, however, still benefits from Siri’s capabilities.
Continued in Part 4.
Base Data for Graph 1.
|Year||Inflation||HC Low||HC Medium||HC High||HC Gamer||Author Home Computers||Author Home Computers and Mobile Devices|
Data for Graph 2 (additional column for data in Graph 1) – these are the initial cost of Watson (3 million as indicated on Wikipedia) declining by 16% per year.