(These are excerpts from my book "Intelligence is not Artificial")
What is the Opposite of the Singularity?
What worries me most is not the rapid increase in machine intelligence but a possible decrease in human intelligence.
The Turing Test is commonly understood as: when can we say that a machine has become as intelligent as humans? But the Turing Test is about humans as much as it is about machines because it can equivalently be formulated as: when can we say that humans have become as stupid as a machine? In other words, there's another way that machines can to pass the Turing Test: make dumber humans. Let's call the Turing Point the point when the machine has become as smart as humans. The Turing Point can be reached because machine intelligence increases to human level or because human intelligence decreases to machine level.
Humans have always become dependent on the tools they invented. For example, when they invented writing, they lost memory skills. On the other hand, they gained a way to store a lot more knowledge and to broadcast it a lot faster. Ditto for all other inventions in history: a skill was lost, a skill was acquired. We cannot replay history backwards and we will never know what the world would be like if humans had not lost those memory skills. Indirectly we assume that the world as it is now is the best that it could have been. In reality, over the centuries the weaker memory skills have been driving an explosion of tools to deal with weak memory. Each tool, in turn, caused the decline of another skill. It is debatable if the invention of writing was worth this long chain of lost skills. This process of "dumbization" has been going on throughout society and it accelerated dramatically ("exploded") with electrical appliances and now with digital devices. The computer caused the decline of calligraphy. Voice
recognition will cause the decline of writing.
In a sense, technology is about giving dumb people the tools to become dumber and still continue to live a happy life. A pessimist can rewrite the entire history of human civilization as the history of making humans dumber and inventing increasingly smarter tools to compensate for their increasing stupidity.
In some cases the skill that is lost may have broader implications. If you always use the smartphone's navigator to find places, your brain does not exercise the part of the brain that knows how to navigate the territory. If we don't explore, we don't learn how to explore. If we don't learn how to explore, we don't grow cognitive maps, and we don't train the brain to create cognitive maps. The cognitive map is a concept introduced in 1948 by Edward Tolman to explain how higher animals orient themselves. Without them, the brain is a diminished organ: it will never learn how to do all the things that are enabled by cognitive maps. If George Lakoff is right and all thinking is rooted in physical metaphors, there are countless thoughts that can happen only to brains that know how to manage cognitive maps. Reading novels and discovering scientific theories may not be possible without cognitive maps.
What can machines do now that they could not do 50 years ago? They are just faster, cheaper and can store larger amounts of information. These factors made them ubiquitous. What could humans do 50 years ago that they cannot do now? Ask your grandparents and the list is very long, from multiplication to orientation, from driving in chaotic traffic to fixing a broken shoe. Or just travel to an underdeveloped country where people still live like your old folks used to live and you will find out how incapable you are of simple actions that are routine for them. When will we see a robot that is capable of crossing a street with no help from the traffic light? It will probably take several decades. When will we get to the point that the average person is no longer capable of crossing a street without help from the traffic light? That day is coming much sooner. Judging
from simple daily chores, one could conclude that human intelligence is not
"exploding" but imploding. Based on the evidence, one can argue that
machines are not getting much smarter (just faster), while humans are getting
dumber; hence very soon we will have machines that are smarter than humans but
not only because machines got smarter.
The age of digital devices is enabling the average person to have all sorts of knowledge at her fingertips. That knowledge originally came from someone who was "intelligent" in whichever field. Now it can be used by just about anybody who is not "intelligent" in that field. This user has no motivation to actually learn it: she can just "use" somebody else's "intelligence". The "intelligence" of the user decreases, not increases (except, of course, for the intelligence on how to operate the devices, but, as devices become easier and easier to use, eventually the only intelligence required will be to press a button to turn the device on). Inevitably, humans are becoming ever more dependent on machines, while machines are becoming less dependent on humans.
I chair/organize/moderate cultural events in the Bay Area and, having been around in the old days of the overhead projectors, i'm incredulous when a speaker cannot give her/his talk because her/his computer does not connect properly to the room's audio/visual equipment and therefore s/he cannot use the prepared slide presentation. For thousands of years humans were perfectly capable of giving a talk without any help from technology. Not anymore, apparently. Can you imagine Socrates telling Plato "Sorry, I can't have a dialogue with you unless you have Powerpoint on your laptop"?
The Turing Test could be a self-fulfilling prophecy: at the same time that we (claim to) build "smarter" machines, we are creating dumber people.
My concern, again, is not for machines that are becoming too intelligent, but for humans who are becoming less intelligent. What might be accelerating is the loss of human skills. Every tool deprives humans of the training they need to maintain a skill (whether arithmetic or orientation) and every interaction with machines requires humans to lower their intelligence to the intelligence of machines (e.g., to press digits on a phone in order to request a service). We can argue forever if the onboard computer of a self-driving car is really "driving", but we know for sure what the effect of self-driving cars will be: to raise a generation of humans that are not capable of driving anymore. Every machine
that replaces a human skill (whether the pocket calculator or the street
navigator) reduces the training that humans get in performing that skill (such
as arithmetic and orientation), and eventually causes humans to lose that
skill. This is an ongoing experiment on the human race that could have a
spectacular result: the first major regression in intelligence in the history
of our species.
To be fair, it is not technology per se that makes us dumber. The very system that produces technology makes us dumber. The first step usually consists in some rules and regulations that simplify and normalize a process, whether serving food at a fast-food chain or inquiring about the balance of your checking account or driving a car. Once those rules and regulations are in place, it gets much easier to replace human skills with technology: the human skills required to
perform those tasks have been reduced dramatically, and, in that sense, humans
have become "dumber" at those tasks. In a sense, technology is often an effect,
not a cause: once the skills required to perform a task have been greatly
downgraded, it is quite natural to replace the human operator with a machine.
Paraphrasing something Bertrand Russell said about Ludwig Wittgenstein, we are weary of thinking and we are building a society that would make such an activity unnecessary. Then, of course, an unthinking machine would equal an unthinking human, not because the machine has become as thinking as the human, but because the human has become as unthinking as the machine.
The society of rules and regulations that humans have built to create order and stability has the side effect of making us "think" less.
The Turing Test can be achieved in two ways: 1. by making machines so intelligent that they will seem human; 2. by making humans so stupid that they will seem
To wit, there could be three stages in human civilization. Stage 1: the coexistence of machine stupidity and human intelligence. Stage 2: the coexistence of machine
intelligence and human intelligence. Stage 3: the coexistence of machine
intelligence and human stupidity.
With all due respect, when i interact with government officials or corporate employees, the idea that these people, trained like monkeys to repeatedly say and do the prescribed routine, will some day be enslaved by intelligent machines does not seem so implausible.
What will "singular" mean in a post-literate and post-arithmetic world?
"Men have become the tools of their tools" (Henry Thoreau, "Walden", 1854).
Back to the Table of Contents
Purchase "Intelligence is not Artificial"