Bits of Books - Books by Title
Humans Need Not Apply:
A Guide to Wealth and Work in the Age of Artificial Intelligence
by Jerry Kaplan
The question of whether machines can think is about as relevant as the question of whether submarines can swim. Whether a website that finds you a date or a robot that cuts your grass will do it the same way you do doesn't matter. It will get the job done quicker, cheaper and more accurately than you possibly can.
New AI systems learn from experience. But unlike humans, not ltd in amount of data they can access and from which they can learn. Imagine how smart you could be if you could read every word as it was published, yet have the time to ponder that as well.
Humans are suckers for the quick win. We want bargains online but the cost doesn't include the gradual closing of local stores or the neighbour that puts out of work.
Programs designed as networks with each node and each layer accepting inputs and producing outputs. The outputs of the lower layers were the inputs of the higher ones. The connections were given numeric 'weights' 1 to 100 which automatically adjusted in response to accuracy assessments.
Apple II had 48K memory - enough to store about 1 second of CD quality music. (Today, a 64 GB phone stores enough for about 12 days.) When Watson won Jeopardy! in 2011, it had access to 200 million pages of data - about 4 terabytes of memory. In 2014, Amazon would sell you 4 terabytes of storage for $150.
Incidentally, when Watson won Jeopardy! it had a couple of major advantages. It wasn't actually 'listening' to the clues read out by the host; they were transmitted instantly when he began talking, so didn't have to wait until finished to get the clue. And could ring in quickly, the millisecond it knew it had a highly probable match, and even before it had established exactly which match it would use.
Huge progress in humanoid robots because advances in machine learning processes combined with inexpensive but sophisticated cameras. Programs can examine pics and videos to rapidly recognize people objects and actions.
Roadkill because critters don't have senses attuned to the threat of 2 ton trucks bearing down on them at speed. In a similar way, we are in peril of becoming roadkill because we lack even the vocabulary to describe the technological changes bearing down on us.
For example, predict that future house-painting will involve a fleet of small drones equipped with a spray gun and a small bag of paint. Contractor films building, uses an app to mark areas to be painted, and sets up a control module to monitor progress. Entire house can be painted in an afternoon, at a fraction of usual cost. And all the technologies required to do this are available today.
Self-drive cars will be less autonomous than they appear today - they will be part of a network of vehicle and roadside sensors. Dept of Transport is developing V2V (vehicle to vehicle) protocols which will utilise the part of the radio spectrum the FCC has allocated for automobiles.
The number of prepaid cell phone cards is an indicator of certain crops in Africa (probably because when farmers see a bigger crop, they are contacting more buyers.)
Could reduce the abuses of HFT (High Frequency Trading)by charging a tiny fee (1/1000th of a cent) for each quote request made, and/or by imposing a one second wait on every trade.
When you load a web page, a furious battle takes place behind the scenes. In the second or so between the time you click on a link and the page actually appears on your screen, hundreds of transactions ricochet around the Internet gathering a huge amount of dteails about your recent behaviour, estimating how likely you are to be influenced by one of the available advertisers, and engaging in a flash auction for the right to try to make an impression on you.
Elaborate ad exchanges have sprung up to handle the price auction for each ad that appears on the pages you load. Even at the lightning speed of the Internet, it isn't practical to run multiple auction rounds. So each bidder selects a single ad from its roster and makes a best offer. The ad exchange awards the opportunity to the highest bidder, but only charges it the price offered by the second-highest bidder (which encourages participants to place their best bid).
Jeff Becos understood early that value of Amazon lay in the data they accumulate about purchasing habits of more than 200 million active buyers, not the profits from transactions with them.
Supermarket loyalty cards offer different discounts based on their assessment of how brand-loyal you appear to be. If you were likely to buy it anyway, why reduce the price?
Our lives will be run by synthetic brains whose goals are to optimise traffic, protect resources and manage health care. But there will also be other systems with less lofty goals - to rent you a hotel room with a street view when there is a river view one available, or to route you through Salt Lake City while keeping direct flight seats available for higher paying last minute travellers.
Trucks will be first vehicles to be completely automated. Highways are well maintained and are far more predictable than suburban streets - fewer random moving obstructions like cyclists and pedestrians. The technology to operate self-driving trucks is available today, and can be retro-fitted to existing trucks at reasonable cost. They can also convoy, or platoon, just inches apart, with significant fuel savings. They can operate 24/7 and have far less accidents due to driver error. These trucks are already operating on private roads such as the 150 at Rio Tinto's Pilbara mine.
(London Times)
In Humans Need Not Apply, Kaplan, another tech entrepreneur turned futurist, divides them into two kinds, 'forged labourers' and 'synthetic intellects'. The former are blue-collar robots that move about in and manipulate the material world; the latter are white-collar data systems, acquiring vast amounts of information and applying them to the world in the form of, say, the ad that pops up on your laptop for just the product you were thinking about.
We used to think (and the recent Channel 4 show Humans still seemed to think) that the scary thing about robots was their acquisition of human-like consciousness. This is irrelevant. The most effective and inhuman thing that robots do is look and learn with perfection. Humans survive by a process of generalisations - judgments. Thanks to the speed and scale of their data acquisition and manipulation, robots thrive on particularities. In work they learn from humans and create their own algorithms for replicating things people do faster and better. Then they sack them.
The reason this is not front and centre of current debate is that neither politicians nor the public understand the power of big numbers. The doubling rate in the processing power of computer chips, as expressed by Moore's Law, is an astounding phenomenon. Ford comes up with a neat demonstration: starting at 5mph, if you doubled the speed of your car every minute, by the 28th minute you would be travelling at hundreds of millions of miles per hour. After 27 years, Moore's Law is still working; the future has already raced beyond our imaginations.
There is a further twist of this knife: inequality. This makes both these books seem, at times, to lack focus since they have two apparently different but equal subjects, economics and robotics. But, in fairness, the two have to be understood together. Both authors agree that the rosy picture of wages tracking productivity and profits, and of 'a rising tide that floats all boats', ceased to be true after 1970. Since then, in America, working and middle incomes have stagnated in spite of soaring productivity and company profits. Meanwhile, the bonus culture has driven executive pay to absurd multiples of workers’ pay, and the financialisation of the economies has created novel but socially useless and ultimately damaging forms of money-making. Here technology and economics link directly. High-frequency trading, in which big deals are done in fractions of a second, creates robotic markets that, for the owners of the robots, are effectively risk free. This is often said to be making money out of money, but, as Kaplan shows, it is really making money out of nothing.
Robotisation is the petrol you chuck on this already blazing fire of inequality. Chillingly, Kaplan warns that the even greater inequality that will result from the rise of machines might be sustainable. Look at ancient Egypt, he says, in which all wealth was concentrated on the pharaoh and a huge proportion of the people were employed building his tomb. We could, in other words, become a serf society, slaves of the machines and employees of the sensationally rich.
Kaplan comes up with a solution in which companies could be given huge tax breaks for improved distribution of assets. Ford imagines such things as new medical professionals, who simply interact with the patient and then hand all the information over to a robot. These people would be less than a doctor but more than a nurse. Such intermediaries are a likely area where jobs will grow.
That these are both voices from within the technological system gives them a high degree of authority. But, though Ford and Kaplan mean well, the chances are that neither will be heard. The fire ahead has so successfully seduced us with its gadgets that we may not realise it is hot until it is too late. For the moment there is no hope that the rise of the robots will not be accompanied by the fall of the humans.
More books on Computers
Books by Title
Books by Author
Books by Topic
Bits of Books To Impress
|