SUPERINTELLIGENCE IN AI

(ARTIFICIAL INTELLIGENCE - The Complete Guide)

https://www.amazon.it/dp/B0CCZV8JC2

The book that captivates me the most for its forward-looking insights on AI is "The Singularity Is Near: When Humans Transcend Biology," a 2005 work by Ray Kurzweil. This book delves into artificial intelligence and the future of humanity, and I find it a fascinating topic to discuss.

Raymond Kurzweil, born on February 12, 1948, is an American inventor and futurist renowned for his contributions to fields such as speech synthesis and voice recognition technology. He has authored several books on AI, transhumanism, and the technological singularity. Kurzweil is a vocal advocate for life-extension technologies and has been deeply involved in exploring the future of nanotechnology, robotics, and biotechnology. In recognition of his contributions, he received the National Medal of Technology and Innovation in 1999, the highest honor in the United States for technological advancement.

The aforementioned book builds upon ideas introduced in Kurzweil’s earlier works, "The Age of Intelligent Machines" (1990) and "The Age of Spiritual Machines" (1999). However, "The Singularity Is Near" introduces a novel concept: the term "Singularity," which I further explore in my notes. This term was popularized by Vernor Vinge in his 1993 essay "The Coming Technological Singularity."

What distinguishes the book is its attempt to characterize evolution as progress through six epochs, each building upon the previous one, with AI as an integral part of this evolution.

1. Epoch 1: Physics and Chemistry – At the beginning of the universe, all information existed at a subatomic level.
2. Epoch 2: Biology and DNA – With the advent of life on Earth, genetic information was stored in DNA molecules, although it took thousands, even millions, of years for organisms to evolve.
3. Epoch 3: Brains – Evolution produced increasingly complex organisms. The emergence of the brain allowed these organisms to change their behavior and learn from past experiences.
4. Epoch 4: Technology – Humans evolved into beings capable of creating technology. We are currently in the final stages of this epoch.
5. Epoch 5: The Merger of Human and Machine Intelligence – Biology and technology will begin to merge, creating superior forms of life and intelligence.
6. Epoch 6: The Universe Wakes Up – This epoch will witness the birth of superintelligence and, with it, human/machine entities expanding into the universe.

Kurzweil also explains that evolutionary progress is exponential due to positive feedback, where the outcomes of one stage are used to create the next. He calls this exponential growth the "Law of Accelerating Returns" and believes it applies to many human-created technologies, such as computer memory, transistors, microprocessors, DNA sequencing, magnetic storage, the internet, and the constant miniaturization of devices. Kurzweil cites historical examples of exponential growth, such as the Human Genome Project and the growth of the internet.

Computational Capacity

A key pillar of Kurzweil’s argument is that reaching the Singularity depends on computational capacity as much as on other factors like the quality of algorithms and understanding of the human brain. Moore's Law predicts that the capacity of integrated circuits will grow exponentially. While Kurzweil acknowledges that the growth of integrated circuits will likely slow by 2020, he is confident that a new paradigm will emerge to continue the exponential growth predicted by his Law of Accelerating Returns. He describes four computing paradigms that preceded integrated circuits: electromechanical, relay, vacuum tube, and transistor. Although it is unknown what technology will follow integrated circuits as the sixth paradigm, Kurzweil believes nanotubes are the most likely alternative among several possibilities, including molecular computing, DNA computing, and quantum computing.

Since Kurzweil believes that computational capacity will continue to grow exponentially long after the end of Moore's Law, he predicts that it will eventually rival the sheer computing power of the human brain. He estimates that the human brain's computing power is around 10^16 calculations per second and 10^13 bits of memory. By 2020, he predicts, $1,000 will buy the computing power of a single brain; by 2045, at the start of the Singularity, that same amount of money will purchase a billion times more power than all human brains combined today. While Kurzweil admits that the exponential trend of increasing computational power will eventually hit a limit, he calculates that this limit will be trillions of times beyond what is necessary for the Singularity.

The Brain

Kurzweil argues that computational capacity alone will not create artificial intelligence. He asserts that the best way to build AI is first to understand human intelligence. The first step is to imagine the brain, to peer inside. Kurzweil claims that imaging technologies such as PET (positron emission tomography) and fMRI (functional magnetic resonance imaging) are increasing exponentially in resolution. He predicts that even greater detail will be achieved in the 2020s (now), when it will be possible to scan the brain from the inside using nanobots. Once the physical structure and connectivity information are known, Kurzweil believes researchers will be able to produce functional models of subcellular components and synapses up to entire regions of the brain. The human brain, according to him, is "a complex hierarchy of complex systems, but not beyond our ability to handle."

In addition to reverse-engineering the brain to understand and emulate it, Kurzweil introduces the idea of "uploading" a specific brain with every mental process, to be placed on a "suitably powerful computational substrate." As mentioned earlier, he writes that general modeling requires 10^16 calculations per second and 10^13 bits of memory, but then explains that uploading requires additional detail, perhaps up to 10^19 cps and 10^18 bits. Kurzweil predicts that the technology to do this will be available by 2040. Rather than instant scanning and conversion to digital form, Kurzweil envisions that humans will likely experience a gradual conversion as parts of their brain can be enhanced with neural implants, slowly increasing their percentage of non-biological intelligence over time.

Consciousness

Kurzweil believes that "no objective test can definitively determine the presence of consciousness." Therefore, he asserts that non-biological intelligences might claim not only to have consciousness but also "the full range of emotional and spiritual experiences that humans claim to have." He believes such claims will generally be accepted in the future.

What is Intelligence? Mistakes Made in AI.

Aside from what Kurzweil writes, artificial intelligence is a perfect example of how science sometimes progresses more slowly than expected. In the first wave of excitement over the invention of computers, it was believed that we finally had the tools to solve the problems of the mind and that within a few years, we would witness a new race of intelligent machines. Today, we are older and wiser. The initial surge of enthusiasm has faded, the computers that impressed us so much decades ago no longer impress us, and we are soberly settling in to understand just how difficult the problems of AI really are.

What is AI? In a sense, it is engineering inspired by biology. We look at animals, we look at humans, and we want to be able to build machines that do what they do. We want machines to be able to learn the way we learn to speak, reason, and eventually gain consciousness. AI is thus engineering, but at this stage, can we also call it science? For example, cognitive science? We like to think it is both engineering and science, but we must admit that the contribution AI has made to cognitive science so far is perhaps weaker than the contributions biology has made to engineering.

The fundamental mistake may have been in addressing the issue of intelligence. What happened was that what people considered intelligence was what impressed them. Our peers are impressed if we can handle complex mathematics or play chess. The ability to walk, on the other hand, doesn't impress anyone. You can't tell your friends "look, I can walk," thinking you'll impress them because your friends can walk too. But what has happened over the past 40-50 years, to the disappointment of everyone who made predictions about where AI would go, is that things like playing chess have turned out to be incredibly easy for computers, while learning to walk has proven incredibly difficult.

It has also been very difficult to endow machines with "common sense," emotions, and those other intangible factors that seem to drive very intelligent human behavior. It seems these may derive more from our long history of interactions with the world and with other humans than from any abstract reasoning and logical deduction. So much so that today we might dismiss chess-playing machines like Deep Blue as "brainless machines." The basic philosophy is that we need a much greater understanding of the animal substrates of human behavior before we can realize the dreams of AI in replicating convincingly rounded human intelligence.

In retrospect, a new vision that makes sense should probably take into account the following: it took 3 billion years of evolution to produce monkeys, and then only another 2 million years or so for languages and all the things that impress us to appear. This is perhaps an indication that once you have the monkey, once you have Homo Erectus, all human abilities can evolve fairly quickly. But the problem, then, is not producing more intelligent monkeys but producing the monkey in the first place. Consequently, there is now a revolution in the field called Artificial Life (AL) and Adaptive Behavior, seeking to reposition AI within a basic philosophy that accepts the need for a much greater understanding of the animal substrates of human behavior before we can realize the dreams of AI in replicating convincingly rounded human intelligence.

So AI is not dead, as some think, but it is reorganizing.

One might say that the AI applications I include in the appendix are "rough AI"; indeed, distinctions are beginning to be made between human-likeThe book that captivates me the most for its visionary exploration of AI is "The Singularity Is Near: When Humans Transcend Biology," written by Ray Kurzweil in 2005. This profound work delves into the future of artificial intelligence and humanity, a topic that I am eager to discuss.

Raymond Kurzweil, born on February 12, 1948, is a distinguished American inventor and futurist, renowned for his pioneering contributions to speech synthesis and voice recognition technology. He has authored several influential books on AI, transhumanism, and the technological singularity. Kurzweil is a prominent advocate for life-extension technologies and is deeply engaged in the future of nanotechnology, robotics, and biotechnology. His contributions to technology were recognized in 1999 when he was awarded the National Medal of Technology and Innovation, the highest honor for technological achievement in the United States.

This book builds upon ideas introduced in Kurzweil's earlier works, "The Age of Intelligent Machines" (1990) and "The Age of Spiritual Machines" (1999). However, in "The Singularity Is Near," Kurzweil embraces the term "Singularity," a concept popularized by Vernor Vinge in his 1993 essay "The Coming Technological Singularity."

What distinguishes this book is its attempt to characterize evolution as a progression through six epochs, each building upon the last, with AI as a crucial element in this ongoing process.

1. Epoch 1: Physics and Chemistry – At the universe's inception, all information existed at the subatomic level.
2. Epoch 2: Biology and DNA – With the dawn of life on Earth, genetic information was stored in DNA molecules, although organisms took millions of years to evolve.
3. Epoch 3: Brains – Evolution produced increasingly complex organisms, and the emergence of the brain allowed these organisms to learn from past experiences.
4. Epoch 4: Technology – Humans evolved into beings capable of creating technology, and we are currently in the final stages of this epoch.
5. Epoch 5: The Merger of Human and Machine Intelligence – Biology and technology begin to merge, creating superior forms of life and intelligence.
6. Epoch 6: The Universe Wakes Up – This epoch foresees the rise of superintelligence, with human-machine entities expanding into the universe.

Kurzweil explains that evolutionary progress is exponential, driven by positive feedback, where the results of one phase create the foundation for the next. He refers to this exponential growth as the "Law of Accelerating Returns" and believes it applies to many human-created technologies, such as computer memory, DNA sequencing, and the internet. Historical examples of exponential growth include the Human Genome Project and the rise of the internet.

Computational Capacity

A cornerstone of Kurzweil’s argument is that computational capacity is as critical as other factors, such as the quality of algorithms and understanding of the human brain, in reaching the Singularity. He argues that while Moore's Law predicts exponential growth in integrated circuit capacity, this growth will likely slow by 2020. However, he is confident that a new paradigm will emerge to continue the exponential growth predicted by his law. Kurzweil discusses four previous computing paradigms that led to integrated circuits: electromechanical, relay, vacuum tube, and transistor. While the technology that will follow integrated circuits is unknown, Kurzweil suggests that nanotubes are the most likely successor among other possibilities like molecular computing and quantum computing.

Kurzweil believes that computational capacity will continue to grow exponentially, eventually rivaling the sheer processing power of the human brain. He estimates that by 2020, $1,000 will buy the computing power of a single brain; by 2045, at the start of the Singularity, that same amount will purchase a billion times more computing power than all human brains combined today.

The Brain

Kurzweil contends that computational capacity alone will not create artificial intelligence. He argues that understanding human intelligence is crucial for building AI, starting with imaging the brain. Technologies like PET and fMRI are improving exponentially in resolution, and Kurzweil predicts that by the 2020s, it will be possible to scan the brain from the inside using nanobots. Once the brain's physical structure and connectivity are understood, researchers will be able to produce functional models of the brain's components. Kurzweil asserts that the human brain, despite its complexity, is not beyond our capability to replicate.

In addition to reverse-engineering the brain, Kurzweil introduces the idea of "uploading" a brain's processes to a "suitably powerful computational substrate." He suggests that this technology will be available by 2040, allowing for a gradual conversion of human intelligence to a digital form through neural implants.

Consciousness

Kurzweil believes that no objective test can definitively determine the presence of consciousness. He argues that non-biological intelligences could claim to have consciousness and experience emotions and spirituality, and such claims will likely be accepted in the future.

The Challenges of AI and the Road Ahead

Kurzweil acknowledges that the progress of AI has been slower than initially predicted. Early enthusiasm for AI was based on the belief that computers would soon solve the mind's mysteries and produce intelligent machines. However, as AI development has shown, tasks like walking, which seem simple, have proven to be incredibly difficult for computers, while more complex tasks like playing chess have been relatively easy. This has led to a reassessment of what intelligence truly is and how it can be replicated in machines.

In summary, while AI has not advanced as quickly as once hoped, it is not dead. Instead, it is reorganizing, with new fields like Artificial Life and Adaptive Behavior seeking to place AI within a broader understanding of human and animal behavior. Kurzweil's vision of the future, with the potential for AI to achieve human-like intelligence, remains a powerful and provocative perspective on what lies ahead.

"ARTIFICIAL INTELLIGENCE - The complete guide" https://www.amazon.it/dp/B0CCZV8JC2

"INTELLIGENZA ARTIFICIALE - La Guida Completa" https://www.amazon.it/dp/B0CP85SPL2