On May 11, 2023 I delivered the keynote presentation at the Octavo Dia conference at the Planetarium of Buenos Aires. The following is an edited transcript of the talk.
If you’ve read the book, “I, Robot”, and watched the movie, “Her”, if you have read other science fiction books or seen science fiction movies, have a mobile device with data connection… Good news! You’re ready for a future that’s quickly approaching. But the reality is that the increasing rate of acceleration of technological change is putting even the experts in a difficult position.
We are a species defined not only by our biology but also by our technology. We must continue to use technology in the best way possible to address the challenges we see around us. What we are experiencing is a superexponential world. An unprecedented period in the history of humanity.
We are seeing trends in software and hardware A progress from mainframes to personal computers, mobile phones and now the world of sensors that form the Internet of things. The problem is that the changes are not happening clearly. They occur in a noisy context, where the margins of error in the data may be overwhelming the data itself, in the signal to noise ratio. That is why those who believe in these trends from the beginning can be seen as very optimistic or as crazy. They are attached to something others don’t see or don’t believe. Another factor is that some ecologists or logistics experts will say, “You’re naive, the trends will stop in a finite world.” But this graph by Ray Kurzweil, futurist and founder of Singularity University, shows generations of successive technologies that form the exponential I have mentioned. Today, the superstructure created by these trends forms the basis of our globally interconnected civilization. It extends throughout the planet, whether in supply chains, scientific communication or geopolitical decisions, forming a virtuous cycle.
The electronic brains of the 1950s were enormous machines that needed great care. The integrated circuits that followed quickly, generation after generation, are now at a point where we are reaching the atomic limits of how small they can be. However, we already have new solutions to keep the self-fulfilling prophecy of Moore’s Law in place. The next generation of computers will be based on exploiting the quantum phenomena.
In the world of software, in a similar way at the beginning, computers didn’t know anything about the world. Every night, when you turned them off, they forgot everything that you had taught them through the perforated cards with which you fed them very carefully. The next day, you had to start again. With time, they became more interactive and one could write on these interactive teletype terminals. The language they used was still very specialized. They then began to understand that in fact to organize the work and assign tasks, we like the analogies we use to represent information, and knowledge, such as documents, folders and the desktop metaphor. That we have an aesthetic sense, that we like color and that we care about the difference between beautiful designs and ugly ones. The computers overcame the geometric limits of our hands. Then they began to see the world and to interpret what they saw, they gained the ability to interpret our movements. They could begin to listen to the world and they began to interpret our words. Conversational interfaces and operating systems are being developed so that they can understand what we say. They are disappearing in the environment, where computers see us, but we do not necessarily see them, and where we talk with them in a natural language. The new generation interfaces are already being tested and are near.
This type of interaction, which some would call fusion between machine and human, becomes more intimate, with computers that describe and understand our thoughts and collaborate with us to model the world and face the challenges through those thoughts. Ancient philosophers wondered what it meant to be conscious. The Golem myths talk about an unconscious matter that is awake. Or also the case of the Mechanical Turk which surprised the royal courts of Europe with its ability to play chess, but turned out to be a device that hid a dwarf instead of being a real robot. It was fake news, an artificial artificial intelligence. In the modern era, artificial intelligence began to develop as a project that was supposed to end in one summer. There is no doubt that the project is taking more time than originally thought. But we are moving forward, there has been a lot of progress. We have discovered that the type of calculations that artificial intelligence needs are the same that our computers carry out when we play video games. We are now using specialized hardware designed specifically for artificial intelligence. Modern platforms on the Internet have generated the enormous amounts of data that artificial intelligence applications need and we are rapidly iterating in the design of our algorithms.
We find more and more powerful and efficient ways to reach our goals. If you read the traditional media, you’ll see that there’s often a game. A new technology is presented and they exaggerate its potential. Then the articles about their failure appear, due to the excessive expectations created by the same press.
For a long time, what AI promised seemed to be a just a dream. Stanford University has always been leading in the development of artificial intelligence. In 2019, they published their first comprehensive report on AI. They gathered data by combining many sources and charting the power of artificial intelligence systems over the years. And they concluded that there were two eras, one until the year 2012, when it was observed that the power of AI doubled every two years, and another since then in which the rate of acceleration had increased and the doubling would occur in eight months. That moment I noted that this was wrong. This kind of change cannot begin and then suddenly stop. A couple of years later it was confirmed that I was right when Jensen Huang, the CEO of NVIDIA, the main producer of GPUs for AI, said at their worldwide conference that he saw AI doubling its power in just four months. I called this phenomenon, the Paradigm of Jolting Technologies. The paradigm of jolting technologies, to describe a new era in which we are observing an increased rate of acceleration. In mathematics, the jolt is the first derivative of acceleration. And we are seeing the same effect that is observed in the rockets. They can continue to increase their acceleration when their engine is at its highest thrust, given that their mass decreases as their propellent is consumed during the ascent. In another example of this, the efforts of consolidated companies such as Google and OpenAI are being disrupted by the jolting developments of open source solutions which allow improvements at a radically lower development cost.
If we already had difficulty adapting to the changes when they were simply exponential, the challenges we face now in times that can already be called super-exponential will be even more difficult with jolting technologies and advanced artificial intelligence.
During the Cold War, the Soviet Union introduced an early warning system against nuclear attacks. However, it was defective and took the reflections of the sun in a formation of clouds as a signal of American missile launches. Only because the officer refused to believe it and did not send the alarm along the chain of command, that we are here now instead of a radioactive desert covering the planet. This story is representative of the reasons why we should not allow the use of autonomous lethal weapons. They should be prohibited, similarly to how chemical or biological weapons are. At the same time, the arrival of autonomous cars is unstoppable, and the regulatory environment is adapting to allow them to circulate on our streets. In the United States or China, in a growing number of cities, now you can call a car to take you somewhere, and there will be no one in the driver’s seat. A clear reason for the rapid spread of autonomous cars is that human beings are not very good at driving. More than one million people die in car accidents worldwide every year due to human errors. We must give the power of decision to our autonomous machines so that they can survive. A pilot should not be able to order an airplane to crash against the side of a mountain.
Now we can see the difficulty and the contradiction. On one hand, we should not give machines an autonomy without human intervention. On the other hand, we cannot afford not to make them autonomous. Therefore, the development of the machine morality is a new and necessary development, which will also lead us to a better understanding of our own morality and ethical decision-making. If you asked a Roman slave if his condition was fair, the answer would have been no. But if you asked him if an alternative was possible, he would have told you that there was no way to create a society without slaves. However, we could abandon slavery thanks to our technology. Human empathy and tolerance are increasing thanks to technology and our ability to decrypt and better understand emotions and the mental states of the others. Our emotional intelligence is increasing.
The cost of human development cannot be seen as avoidable thanks to higher efficiency of our increasingly automated systems. And the social contract today is simple. Your value for society is equal to your economic production. If you lose your job, you are worth zero.
100 years ago, the word robot entered our global vocabulary based on the theatrical work and novel RUR by Karel Čapek. The Hollywood distopias to which we are accustomed when the future is represented is not how things are going to develop. The enemies are not the smart machines. The enemies are the dumb machines that together with often dumb people cause harm. And the coming humanoid robots will for the first time decouple the economic production of the countries from their population. It has become urgent and necessary to have a deep and inclusive conversation about our societies, a real refunding that must lead to a new social contract.
What is the future of humans in a world where we will coexist with machines? Some of us are radically experimenting with these technologies. What is the future of machines in a world where we will co-evolve with them? Some of us wonder what will happen when humans and machines merge? What society will be designed by the type of physical and cognitive enhancements that will we see? And what is the form of society that will arise from it? In today’s world, many of us feel that we are already extending beyond the limits of our adaptability. In the process, we all end up maladapted to a certain degree and we need help.
The luminaries that feel everything, the full spectrum of emotions and are fully self-conscious, are the opposite of the fictional characters of the dead zombies. We can aspire to be all luminaries in our search to understand the world and to understand each other. The technological change is now reaching a phase transition. We are about to meet with alien minds. For a long time we have thought that human intelligence represented a level that could not be surpassed. But soon we will share the world with other intelligences, potentially very different from ours.
Once we have general artificial intelligence, their degrees of freedom will greatly exceed those of humans. There is no clarity regarding the mechanisms for coordinating and negotiating with them our mutual objectives, access to resources, and the implementation of decisions. In the same way as today it is very difficult to be a full member of society if you don’t know how to read and write, it’s possible that we will have to merge deeply with technology only to keep up. We are ready to redefine what it means to be human.
Our self-consciousness is matter that is waking up in a process that took 13 billion years. Androids will dream of the electric sheep. And together we embrace the gradient of values that implement non-zero sum game that we call life. We have a universe to explore and to wake up in front of us.