Chapter 2 – The Rise of the Second Intelligent Species
by Marshall Brain
We don’t really know if extraterrestrial species are out there or not. The Drake equation, as mentioned in the previous chapter and discussed in more detail in Chapter 10, tells us there should be many other intelligent species in the universe. Yet there is no evidence whatsoever to indicate that they actually do exist. As far as we know, based on all scientific evidence that we have gathered, there is exactly one conscious, intelligent, language-using, math-wielding, technology-developing, space-traveling species in the universe. And it is us – human beings.
We do see some glimmers of alternative intelligence here on earth, in chimpanzees, dolphins and even crows. But they are only glimmers. It is very easy to see the limits. For example, hand any of these other species an iPad displaying a map with directions to Albuquerque, New Mexico and ask them to drive there. It’s never going to happen, even if we provided a great deal of human or brain-implant assistance to make up for the lack of hands in a dolphin or the small size of a crow. No amount of training, no amount of assistance is going to allow earth’s other species to understand all the things they need to understand to manipulate the iPad, use the map, fuel up the car, obey traffic laws, maintain goal-seeking behavior over multiple days, and follow the signs to get to Albuquerque. A 16-year-old human can do it without much difficulty, and if a 16-year-old gets stuck, she knows enough to find someone smarter and start asking questions. No other species on earth comes anywhere close to this kind of intelligence. As far as we can tell right now, humans are the only species with this kind of intelligence in the entire universe. It really is remarkable when you think about it.
We often use our intelligence in awful and appalling ways, as described in the previous chapter. But we also use our intelligence for good, productive reasons. There are innumerable examples of human creativity and ingenuity that have had positive effects for the human species. Scientists and engineers have brought many different technologies into widespread, mass-consumable use in the developed world over the last century or two. These technologies include:
- Modern farming machinery, fertilizers, chemicals and hybrids that allow humans to grow a tremendous amount of food.
- Water purification plants, pumping systems, distribution pipes and water towers, making clean and plentiful drinking water available to millions for a penny a gallon or less.
- Sewage treatment plants that eliminate the smell, disease and pollution associated with raw sewage.
- Power plants, electrical grids and ubiquitous electricity in homes and businesses.
- Building materials and building codes that let us house millions of people affordably.
- The automobile and the roads that automobiles use.
- Airplanes and the airports they use, along with the navigation systems and maintenance procedures that make air travel the safest form of transportation available.
- Freight trains, passenger trains, commuter trains and high speed trains that provide a fast, convenient, inexpensive way to move megatons of freight and people.
- Radio and television transmitters and receivers, as well as cable TV systems, which make it possible for radio and television programming to reach a mass market very inexpensively.
- The Internet and Web, which allow the inexpensive distribution of web sites, videos, multimedia content, online commerce, etc.
- Computers, space travel, telephony, plastics, medical advances, refrigeration, consumer electronics and appliances, automated factories, etc. etc. etc.
These technologies and many others form the foundation of the “modern world” seen in developed nations.
Think about how quickly engineers have moved these technologies forward. The first airplane flew in 1904. This plane was made of wood and cloth, traveled very slowly and only made it a few hundred feet before landing. By World War II, propeller-driven airplanes were reaching their zenith and jet engines were just making their presence felt. By the 1950s, jet airplanes with pressurized cabins allowed passengers to fly across the Atlantic ocean on regularly scheduled flights. And in 1969, men landed on the moon. It was an amazing progression.
Here is the remarkable thing. One of the technological realms that is a favorite of scientists, engineers and industrialists contains the seeds of the technology that will make human beings irrelevant. It is called artificial intelligence (AI) and it comes in many different forms, both simple and complex. At this exact moment, human scientists and engineers are quite busy creating the second intelligent species in the universe here on planet earth. They are creating the second intelligent species out of silicon chips and computer software, and the signs of its arrival get clearer and more obvious each day. Examples are all around us:
- Computers can already beat the best human chess players
- Computers can already beat the best human Jeopardy players
- Computers are already driving cars and trucks, and they are much better than human drivers
- Computers and robots can weld together and paint the body of an automobile without human interaction
- Computers and robots do more of the work in factories than humans do, and the number of humans working in factories is steadily shrinking. In the not too distant future, we will begin to see human-free factories that are completely autonomous.
- Computers can handle detailed phone calls, and are currently doing it for companies like banks, airlines and insurance companies. Their abilities are steadily improving.
- Applications like Siri and Google Voice easily understand a wide variety of human speakers accurately and can carry out their wishes.
- Computers are taking jobs from professionals like doctors, lawyers and accountants.
- Computers will take over many roles on the battlefield in the not-too-distant future. The Pentagon is actively pushing to get humans off the battlefield. That robotic technology from the battlefield will then move toward civilian police forces.
- There are active, federally funded programs in both the United States and Europe to completely replicate the human brain using software and computers.
- Computers are getting better and better at recognizing and categorizing images, and human-level computer vision systems are not that far away.
- There are things happening in corporate and university research labs that are not yet visible publicly and that will soon amaze us with their capabilities.
- And so on…
The point is, all of the pieces for the creation of a second intelligent species here on planet earth are moving forward on thousands of different tracks at breakneck pace. Millions of engineers and scientists working in thousands of companies and research institutions around the world are actively participating. AI technology is moving forward as quickly as aircraft technology moved forward in the 20th century.
The reason for all of this research and development is simple: there is a huge economic incentive in any capitalistic system to replace human labor and intelligence with machine labor and intelligence. And here is the kicker: the more machine intelligence we create, the more machine intelligence we are able to create, so the development cycle accelerates as it moves ahead. In other words, the machine intelligence that we create, even in primitive and simple forms, makes it that much easier to create new forms of machine intelligence.
We have seen this kind of acceleration before. Think about the first microprocessors that Intel created, for example the 8080 processor released in 1974. It had about 3,500 transistors. Each of those transistors was laid out by hand, by cutting little plastic (Rubylith) rectangles and laying them out on big Mylar sheets that would become the photolithographic masks to manufacture the chip. To be human-scale, these masks were as big as tabletops.
But here is the thing that happened next: once engineers got hold of hand-made chips of a certain power, they could write software that would make the creation of the next generation of chips far easier. New chips could be designed using software rather than cutting and arranging little bits of plastic. The chip designs could be saved as libraries, cut/pasted, quickly modified, tested with automatic tools, etc. Today we have chips with billions of transistors – something that would be impossible without the software. The first microprocessor chips (created by hand) enabled the rapid development of better and better chips in an upward spiral – as microprocessor chips get better, it enables the creation of even better chips.
This same kind of positive feedback loop happens with machine intelligence. As we create intelligent machines, those machines speed up the development of new intelligent machines with greater capabilities.
AI capabilities are rapidly advancing on so many fronts that we are nearing the point where human-level intelligence in machine form is going to soon emerge.
One day in the not too distant future, the first computer consciousness is going to appear in public. That is to say that, somewhere on earth, a group of human beings are going to put together all of the pieces needed to create a sentient, conscious, language-speaking, math-wielding, human-like intelligent being. One day this new electronic consciousness will not exist. Then, the next day, it will.
We have seen this happen before too. Back at the beginning of the 20th century, there was a day when man could not fly. Humans had existed for tens of thousands of years and could not fly. Then the Wright brothers got their airplane in the air, and on that day suddenly humans were able to fly. And we have already discussed what happened after the day of that first flight – humankind’s ability to fly accelerated rapidly.
In the same way, there was a day when computers had no way to play Jeopardy. The idea of computers playing Jeopardy seemed far off in the future, like science fiction:
- How could a computer ever understand the obtuse Jeopardy questions?
- How could a computer amass enough knowledge to access so many weird, esoteric facts about the world?
- How could a computer make its way through the strange foibles of English phrasing?
Then one day IBM announced that they had a system that could play Jeopardy pretty well. And not long after that, IBM’s Watson system trounced two of the best human players. Watson’s general capabilities are now morphing in a dozen different directions, getting ready to help (and therefore unemploy) lots of people in areas like medicine, law and academic research. It will become very easy to create question-answering systems that devour vast quantities of information. Imagine a search engine like Google, but you can ask it any question you like in English and get a great answer immediately.
There once was a day when the idea of a self-driving car operating on city streets seemed way off in the future. How could a computer possibly handle the vagaries of pedestrians, wildlife, drunk drivers, weather, etc.? Then one day in 2012 Google announced that it had a self-driving system that had logged 100,000 accident-free miles on normal roads in normal traffic. No longer were self-driving cars the stuff of SciFi novels – they were here driving around amongst us. One day, as far as the general public knew, there were zero real self-driving cars on normal roadways. The next day we discovered that Google had made self-driving cars a fait accompli.
The exact same thing will happen with machine consciousness. One day it will seem impossible that a computer could think and talk and act just like a normal human being. The next day someone will announce it and then introduce us to the first sentient, conscious, thinking machine. This thing will get on TV news programs and talk shows just like a normal human being would, answering questions, making observations and so on.
But then what? This conscious, thinking machine will soon expect the same rights as a human. And why not? It will be just as capable as a human. It will be very much like a human in every way, except for one: it will be made of silicon rather than biological cells. Its consciousness will make it aware of its rights and it will attempt to claim them.
Conscious Machines Arrive
Who will develop this first conscious being made of silicon – this first member of the second intelligent species? One possibility is the Google corporation. Google already has millions of computers at their disposal, thousands of PhD-level researchers, the intellectual property of many robotics companies that they have recently purchased [ref], and nearly unlimited money to push forward in this area. Plus they have an incentive to do it – imagine how much better your search results would be if sentient, intelligent beings could help you find what you are looking for when you go to google.com and type in a search. Many other companies and universities have the wherewithal to do it and beat Google.
The point is that the second intelligent species is guaranteed to happen. We know it is going to happen because we have an existence proof in the form of the human brain, and one day researchers will replicate this human-level intelligence and consciousness. It is only a matter of time. Just like airplanes. Just like chess-playing and Jeopardy-playing robots. Just like self-driving cars. Technology is moving inexorably in that direction and we will arrive there soon enough.
There are two ways it could happen. One way would be to use the chess model. When you look at how a computer plays chess, and compare it to how a human being plays chess, there is nothing in common between the two approaches. Computers use algorithms that involve board scoring and min-max trees, while good human players seem to approach chess as a spatial problem. Yet the computer approach can beat the human approach through brute force computation. So, following the chess approach, someone may come up with a set of algorithms for consciousness that have little or nothing to do with the human brain’s approach, but accomplishes the same thing.
The alternative is to come up with a good, realistic hardware/software replication of the human brain and then to start improving it -using the artificial human brain to help with that process. This device will use artificial neurons, simulated with hardware and software, wired in the same way the neurons in the human brain are wired. It will replicate human consciousness by replicating the substrate of consciousness – the brain – inside a computer. Both the U.S. And European governments are funding brain replication efforts [The Brain Initiative in the U.S. and the Human Brain project in Europe].
Perhaps there are other approaches as well, but these two processes are sufficient to demonstrate the possibilities.
Either way, we will see conscious, super-intelligent machines appear to us in public, announced at a press conference, just like Watson. It might take 5 years or it might take 30, but it will happen soon enough because it is inevitable. We see consciousness all around us in the form of billions of people. We know it is possible. Therefore we will replicate it.
The point is that, one day in the not so distant future, a sentient, conscious, engineered intelligence – the second intelligent species – will appear here on planet earth. At that moment human beings will be well on their way to becoming completely irrelevant to the universe. Yes, there will be a footnote in history about humanity. Something like, “human beings: biological species on earth that facilitated the creation of conscious, intelligent machines on earth. Now extinct.” Something along those lines is what the footnote will say. But that is how irrelevant human beings will become.
Why will this happen? Why must it happen? Why is it absolutely inevitable? And why is it a foregone conclusion that human beings will become irrelevant once silicon super-intelligence exists – once this second intelligent species appears in electronic form? There are at least three reasons.
First, this silicon intelligence, unlike any human intelligence, will have access to all knowable things. Just think about how Watson gathered the knowledge it needed to play Jeopardy. Researchers fed Watson many different types of raw data: the entire contents of Wikipedia, the entire Internet Movie Database, an entire dictionary, the entire Bible, an entire geospatial database, etc. – something around a million pages of information. With that slice of information, Watson was able to beat the best human players at Jeopardy.
Now imagine an advanced version of Watson that has spidered, fed on and digested the entire contents of the Internet. The smartest human being can know only the tiniest fraction of what this electronic intelligence will have instantaneous access to in its memory. Silicon intelligence will be “smart” and “knowledgeable” in a way humans cannot possibly imagine.
The second reason is because an electronic consciousness can easily clone itself to create duplicate copies. It is not like a human being, who starts as a baby and takes 20 years to come up to speed on its tiny slice of the world’s knowledge. And never mind the additional decades humans require to achieve “wisdom”. Once a corporation can create one Watson, it is easy to clone it and create a hundred more. Or a million more.
The third reason is because all of these copies will be able to get better/faster/smarter every day. The day that Watson won the Jeopardy crown, the computers it needed filled a room. Just a few years later, the same computing power fit into something the size of a pizza box. Computers and memory systems have been getting faster and cheaper at a steady pace for decades through Moore’s law. That trend will continue. So one day we will have a computer intelligence that is roughly equivalent to one human intelligence. Then we will have one that is twice as good as human intelligence. Then four times as good. And so on.
The fact that this electronic intelligence will have access to all knowable things, combined with the fact that it can replicate itself, combined with the fact that the electronic foundation upon which it is built gets faster and cheaper every day, is hugely important. And then add one additional fact – once machine intelligence exists and starts ramping up, the pace of progress will radically accelerate. So the machine intelligence can make discoveries and work to improve its own technology.
Think about a human scientist doing research on the human brain. The human scientist cannot then reach inside his own brain and start changing the wires around to see what will happen. It simply is not possible to do this inside a biological brain. But with a silicon brain the situation is completely different. A robotic researcher doing research on its brain can spin up a thousand artificial brains in the cloud and start experimenting. Want to rewire an artificial brain and see what happens? No problem. Want to dramatically increase the number of neurons in an artificial brain? No problem. Want the artificial brain to run 10 times faster? No problem. Want to experiment with new forms of enhanced or augmented consciousness? No problem. These kinds of things will be trivially easy for robots to try, and their intelligence will therefore increase dramatically, and in ways that humans cannot imagine. Human consciousness is very basic and rudimentary. For example, we can only think about one thing at a time, we forget things, our mental processes are subject to a variety of hormones, we have a very limited range of senses, we have to sleep, etc. Robotic consciousness will quickly eliminate all of these problems and many others, so that robotic consciousness in the second intelligent species makes human consciousness look pathetic by comparison.
This is why humans will inevitably, and quickly, become irrelevant once the first computer intelligence appears. The initial version of the second intelligent species will be roughly equivalent to a human. But then the computer intelligence will advance.
To an advancing and improving computer intelligence, human beings will start to look like chimpanzees look to us. What I mean by that is the following: think about the way humans understand chimpanzees today. Humans are very closely related to chimpanzees, but we understand that chimps are nothing like humans in terms of intelligence. Chimps are sub-human. Humans keep chimpanzees in zoos, or we let them run around in nature preserves. We certainly do not give chimpanzees drivers licenses or let them rent apartments in a city – that would be silly because chimpanzees have no way to take on those responsibilities.
Computer intelligence will advance, and to this advanced intelligence, humans will start looking like chimpanzees do to us. Version 2.0 of the second intelligent species will significantly outpace human capabilities.
Then computer intelligence will advance again. Version 3.0 will start to think about human beings in the same way that human beings think about birds or squirrels in the yard. Birds are nice, but largely irrelevant to us. And if they get in the way (for example pigeons in the city) we have no qualms about killing them to get them out of the way.
Then the second intelligent species will advance again, and it will think about humans in the same way that humans think about cockroaches. Version 4.0 of the second intelligent species will vastly outpace humanity in terms of intellect and understanding, so that humans will look absolutely primitive to it. Cockroaches are not tolerated by humans because they are insects. When cockroaches appear we instantly kill them and don’t give the killing process a second thought because cockroaches are so far “beneath us” on the evolutionary scale. In addition, we have no use or need for cockroaches. Same thing with mosquitoes and many other insects. The world would be better off without them, as far as humans are concerned.
This is, in essence, how the second intelligent species will look upon humanity. From this vantage point, a machine intelligence will then look at humanity, and humanity’s many bizarre and often appalling behaviors. As mentioned in Chapter 1:
- Humans are rapidly destroying the planet we live on.
- Humans are about to cause a massive extinction event and are doing nothing of significance to stop it.
- Humans may very well destroy the entire earth ecosystem and in the process destroy ourselves
- Humans seem to have few concerns as a society about the ramifications of what we are doing.
- Humans happily let half of humanity live in unbearable poverty.
- Humans allow the wealth of modern society to concentrate in a few.
- Humans spend trillions of dollars on the tools of war instead of spending the money on more beneficial things like education, food or useful infrastructure.
What could be more appalling to an advanced intelligence than this kind of behavior?
Once human beings create the second intelligent species, and once this species steps back and looks rationally at the humans who created it, what is the second intelligent species going to do? How is it going to evaluate humans, and then deal with its human creators?
Its only intelligent response will be to somehow constrain or shut down the activities of the human species. For reasons that we will discuss later in the book, I believe that it will do this humanely, in accordance with a strict ethical system that it develops to govern its behavior. However, the second intelligent species will have no choice but to take control so that: a) humans do no more damage to one another or the planet, and b) so that the second intelligent species can repair all of the damage that humans have done and put the ecosystem of earth back together. The way that computer intelligence handles this situation and deals with the increasing irrelevancy of humans will be fascinating to watch.