Nick Bostrom, Superintelligence and Simulation Hypothesis. Or do we need Philosophy?

I had the great pleasure of being at lunch with a scientist this week. A talented physicist and also a data scientist. We talked about various things in the context of the application of data science to science and the development of science in general. At one point we touched on the topic of what science is in general and whether it is philosophy.

According to my interlocutor, philosophy ended somewhere at the end of the 18th century. And it was replaced by physics and mathematics. Maybe ethics remained as an independent science.

So I began to wonder whether we need philosophy in the ChatGPT era. My feelings are mixed.

I have been trying to finish the book "Superintelligence" by Nick Bostrom for a long time. And I finally managed this weekend. Nick Bostrom is a Swedish philosopher, futurist, and influential figure in the field of artificial intelligence (AI) and existential risk.?

The book explores the potential and risks associated with the development of artificial general intelligence (AGI). Bostrom delves into the implications of AGI surpassing human intelligence and highlights the need for careful planning and preparation to ensure a positive outcome for humanity.?

The book begins by laying the groundwork for the concept of AGI and its transformative potential. Bostrom convincingly argues that AGI is not just a distant possibility but an inevitable outcome of technological progress. He skillfully explains how AGI differs from narrow AI and why its emergence could lead to exponential advancements, potentially surpassing human intelligence in ways we can scarcely comprehend.

The heart of "Superintelligence" lies in its exploration of the risks associated with AGI development. Bostrom presents three key concerns: the control problem, value alignment problem, and the competitive race for AGI. He delves into each of these issues, providing a comprehensive analysis of the challenges that AI developers and policymakers will face.?

His emphasis on the importance of value alignment and safety measures resonates deeply, urging us to address these critical issues well in advance.

Throughout the book, Bostrom raises profound questions about the nature of intelligence, consciousness, and humanity's place in the cosmos. He probes the ethical implications of creating superintelligent entities, forcing readers to grapple with the responsibility and potential consequences of playing "creator" to beings that could surpass us in every way.

The book was written in 2015, so in light of the pace of AI development it is already years old. What is striking, however, is how much of Bostrom's musings have become wildly topical in the context of the generative AI explosion.?

What I liked about the book, and why I think the philosophy isn't dead after all, is the methodology for analyzing the risks of artificial intelligence. And for me as an ordinary bread eater, philosophy makes sense here as a tool to organize thinking.

The book is a great summary of AI history and AI development, too.

What I didn't like, on the other hand, was the author's "departing" into divagations that are, in my opinion, again a non-scientist, not scientifically grounded in anything, and reduce the discussion of AI to the realm of science fiction books and movies.

Bostrom creates a picture at one point when the world is ruled by AI through the emulation of humans. Humans are deprived of capital, which AI takes over, and become slaves or serfs who live for the proverbial "bowl of rice." A hierarchy is created among the emulations and a social system is formed....

On the other hand, in such vision we have incredible advances in medicine, we probably know how to create feelings of happiness in people with the help of AI. This vision, monstrous to me, is called "Disneyland without children" by Bostrom.

Perhaps the book is written provocatively so that such a "Disneyland without children" will never be created. But does the purely theoretical creation of black visions make sense?

Nick Bostrom is also one of the creators of the simulation hypothesis (in the links interesting interviews with Lex Fridman and Joe Rogan).

The hypothesis suggests that we might be living inside a computer simulation created by an advanced civilization with immensely powerful computing capabilities.

The core of the Simulation Hypothesis can be summarized in three propositions:

• It is possible for a technologically advanced civilization to create highly realistic and sophisticated computer simulations that contain conscious beings. These simulations could be so realistic that the simulated beings within them would have no way of knowing they are in a simulation.
• If a technologically advanced civilization were to develop, it might reach a stage where it possesses the ability to create vast numbers of such simulations, each containing complex and conscious entities.
• Given the potential for the creation of numerous simulations, the number of simulated conscious entities could vastly outnumber the number of real, non-simulated conscious beings in the universe.

Based on these propositions, Bostrom argues that it is statistically more likely that we are living in a simulated reality rather than a "base reality". In other words, the hypothesis suggests that our reality is one of many simulated worlds, and what we perceive as the physical universe is a simulated construct.

It is important to note that the Simulation Hypothesis is not presented as a definitive theory but rather as a philosophical argument that explores the implications of a postulated future scenario. Bostrom himself does not claim to have evidence for or against the hypothesis; instead, he argues that if certain technological advancements are possible and that future civilizations choose to create simulations, then the odds of us living in a simulation might be significant.

As an ordinary bread eater, I take it as an intellectual game to provoke us to think, in the context of VR creation. Bostrom himself in interviews dismisses his hypothesis as an intellectual experiment. So, if you have good weather on a Sunday afternoon I recommend going for a summer walk. I may not have many intellectual arguments, but at my place 29 centigrade is warming up nicely which convinces me we are not living in a simulation. Some philosophy is intellectually stimulating, but one needs to dose it in right proportion.