Witness the fascinating world of AI, with Toby Walsh, one of the world’s leading researchers in Artificial Intelligence, in his latest book, 2062 – The World that AI made. (Published by Speaking Tiger, 2020)
THE MACHINE ADVANTAGE
To understand why Homo sapiens is set to be replaced, you need to understand the many advantages computers have over humans and that the digital world has over the analog. Co- learning is one important advantage, but let’s look at some others.
The first is that computers can have a much more expansive memory capacity than humans. Everything we remember has to be stored in our bony craniums. Indeed, we already pay a significant price for having heads as big as they are. Until recently, childbirth was one of the major causes of death for women. And the size of the birth canal limits us from having bigger heads still.
Computers have no such limits. We can simply add more storage. The second advantage is that computers can work at much higher speeds than humans. The brain works at under 100 Hz as neurons take over 1/100th of a second each to fire. Our brains are chemical as well as electrical, which slows them down further. It takes time for chemicals to cross nerve boundaries, and for chemical reactions to take place. Computers, on the other hand, are bound only by the laws of physics. The speed of computers has gone up from 5 MHz in 1981 (that is, able to execute five instructions every millionth of a second) to around 5 GHz today (able to execute five instructions every billionth of a second). Of course, speed alone is not a very good measure today of performance. Raw computer speed has not increased much lately. Instead, computers now get faster by doing more at once. Just like the human brain, computers now execute several instructions simultaneously. Nevertheless, there remains a fundamental speed advantage of silicon over biology.
The third advantage that machines have over humans is that, unlike computers, humans have a limited power supply. Our brains use around 20 watts of the 100 watts an adult body produces. The evolutionary advantage of being smart justifies investing so much of our body’s limited power in the brain, but we don’t have any spare power to invest in more thinking. By comparison, an average laptop can draw up to 60 watts of power.
The brain works at under 100 Hz as neurons take over 1/100th of a second each to fire. Our brains are chemical as well as electrical, which slows them down further. It takes time for chemicals to cross nerve boundaries, and for chemical reactions to take place. Computers, on the other hand, are bound only by the laws of physics.
And if you want more power (or computation), you can simply run jobs in the cloud. The 7 billion human brains on the planet collectively consume around 14 gigawatts of power. By comparison, computing worldwide already uses more than ten times that amount of power. In fact, computing today accounts for 10 per cent of the world’s electricity use, or over 200 gigawatts. And this number will only grow.
The fourth advantage that computers have over humans is that humans need rest and sleep. Computers can work 24/7 and never tire. As we saw earlier, AlphaGo became so good at Go because it played more games of Go than any human ever could. Of course, for humans, sleep likely serves various purposes beyond resting and recovering strength. It may help us update our memories, for instance, and tackle problems subliminally. Perhaps computers might benefit in the same way? We could choose to program them to sleep for a day every so often.
The fifth advantage that computers have over humans is that humans are forgetful, while computers aren’t. Think how often you’ve wasted time looking for a lost object. Or forgotten a birthday. Forgetting can sometimes be useful, helping us ignore irrelevant details but it’s a trivial matter to program computers to forget.
The sixth advantage is that humans can be blinded by their emotions. Computers today have no emotions and so cannot be blinded. On the other hand, emotions play an important role in our lives, and often influence our decision- making in positive ways too. It seems likely, therefore, that they have evolutionary value. We may choose to give emotions to computers in the future. In chapter three, I’ll discuss this in more detail, along with other challenging topics such as consciousness.
The seventh advantage computers have over humans is one we have explored already: that humans are limited in how we can share our knowledge and skills, while computers are not. Any computer can run the code of any other computer. When one computer learns to translate from Mandarin to English, we can give that ability to every computer. When one computer learns to diagnose melanoma, we can give that skill to every computer. Computers are the ultimate co- learners.
Humans are, in reality, rather poor decision- makers. We’ve evolved to be good enough to survive, but this is far from optimal. We’re terrible at calculating exact probabilities, for instance. We would never buy lottery tickets if we were better at this.Tweet
The eight advantage is that humans are, in reality, rather poor decision- makers. We’ve evolved to be good enough to survive, but this is far from optimal. We’re terrible at calculating exact probabilities, for instance. We would never buy lottery tickets if we were better at this. But we can program computers to be optimal. The field of behavioural economics studies our suboptimal decisions. For instance, we are often driven not to maximise our profits but to avoid losses. Behavioural economists call this loss aversion. There are many other examples of suboptimal behaviour. Many of us fear flying when it is the drive to the airport, we should be more worried about. We know we should lose a few pounds, but that jam donut is just so tasty.
Of course, it’s not all one way. Computers aren’t better than us in all respects. Humans have a couple of major advantages over computers. Our brains are still more complex than even the largest supercomputers. We are quick learners, and have astounding creativity, emotional intelligence and social empathy. But I very much doubt that we will maintain these advantages over computers in the long term. We already have some evidence that computers can be creative, as well as possess emotional intelligence and empathy. In the long term, Homo sapiens has little hope in a race against the machines.
Who, then, is this Homo digitalis, this even more remarkable species that will replace us?
A species is defined both by what it is physically and where it acts geographically. In the case of Homo digitalis, both what it is and where it acts will be increasingly digital. Homo digitalis will start out as the digital version of ourselves. As computers become smarter, we will outsource more and more of our thinking to them. These digital entities will no longer be held back by our complex, messy and somewhat limited brains. We will scape the limitations of bodies that need to rest and sleep, and that eventually decay and die. We will no longer be limited to observing and acting in one place at a time. We will be everywhere simultaneously.
By augmenting our brains digitally, Homo digitalis will be far smarter than Homo sapiens. Increasingly, the distinction between what we think and what is thought in the AI cloud will be hard to distinguish. Homo digitalis will transcend our physical self, and will be both biological and digital. We will live both in our brains and in the larger digital space.
In fact, for much of the time, homo digitalis will no longer have to be part of the slow, messy and dangerous analog world. Increasingly, we will live and act in a purely digital world. After a century of climate change, financial crises and terrorism, this digital world will be a welcoming, organised and well- ordered place. There will be none of the uncertainty that makes living on Earth so painful at times. There will be no earthquakes or landslides. No plagues. Everything will follow precise and fair rules. Homo digitalis will be the master of this digital universe. We will, in some sense, have become gods of this digital space. That’s the optimistic outcome – because we get to build this digital future. We really are gods in this sense. And we can ensure that this digital future is fair, just and beautiful. Or we can let the current forces shaping our planet define what it is, and allow it to be full of inequality, injustice and suffering. We get to choose. And we start to make these choices today.
The future is not inevitable. It is the product of the decisions we make today. But it seems likely to me that we are at a critical juncture. There are many forces pushing us towards a slippery slope, one that leads to a very troublesome and disturbing world.
The future is not inevitable. It is the product of the decisions we make today.Tweet
We have the chance now to make some choices that will save us from this end, and guide us towards a brighter digital future. Some of these choices will be easy and cheap, others difficult and costly. They may require vision, leadership, selflessness, even sacrifice. We have been very lucky. We’ve had the run of our planet this amazing blue- green dot, revolving around a rather typical star on a minor spiral arm of the Milky Way – for the last few hundred thousand years. We owe it to our grandchildren – who will after all be of this new species Homo digitalis – to get the next few decades right.
Excerpted from 2062: The World That AI Made by Toby Walsh. Published by Speaking Tiger Books, 2020.
About the Author
Toby Walsh is one of the world’s leading researchers in Artificial Intelligence. He is a Professor of Artificial Intelligence at the University of New South Wales and leads a research group at Data61, Australia’s Centre of Excellence for ICT Research. He has been elected a fellow of the Association for the Advancement of AI for his contributions to AI research, and has won the prestigious Humboldt research award. He has previously held research positions in England, Scotland, France, Germany, Italy, Ireland and Sweden.