The use of AI tools can be both wise and unwise. On the one hand, AI can aid us in detecting criminals, but on the other hand, it can be misused to spy on and enforce undemocratic surveillance. AI has the potential to evaluate complex datasets and reveal patterns that we may not have been able to see on our own. However, it can also trap us in information bubbles and echo chambers, which can distort our perspective of the world and hinder our ability to form responsible opinions on important issues such as vaccinations and climate change. It is clear that while AI can be a great servant, it also has the potential to become a bad master.

In recent times, artificial intelligence and machine learning have become ubiquitous topics in public discourse. Their capabilities, at times, seem like they belong in a science-fiction novel. AI can converse with us in natural language, write poems, stories, and books, analyze complex texts, and generate stunning images based on instructions. And this is just the beginning. An advanced language model, GPT-4, has now been made available to a select number of users, overcoming some of the shortcomings of its predecessor. It has integrated new features and is equipped to handle even more complex tasks. Moreover, it is just the first in a series of advancements that will be followed by other, even more impressive cognitive tools that can solve an increasingly wider range of problems.

However, AI's astonishing capabilities are not the sole reason for its widespread discussion. Recently, a group of prominent experts in AI research and related fields have published an open letter calling for a halt in AI development for six months. The letter warns of potential risks associated with such advanced AI systems and calls for a thorough evaluation of the associated risks. The six-month period should be used to carefully think about and clarify the potential dangers of using these systems and to propose mechanisms to minimize them. Some experts have even revived the idea of AI being an existential risk, perhaps even the greatest one. In a worst-case scenario, AI could be humanity's last invention, leaving nothing meaningful behind. This outcome does not necessitate AI wanting to exterminate humans; it could be sufficient for AI to be fixated on accomplishing a particular task, with humans inadvertently getting in the way. We humans typically don't pay much attention to creatures we consider less than perfect from our anthropocentric perspective. When constructing a new dam, for instance, we seldom consider the fate of an ant colony.

Are these concerns regarding artificial intelligence truly justified? Should we indeed put a halt to its development, possibly for six months, and take the time to reflect on the ethical and societal risks and challenges that it poses? 

In both cases, the answer to these questions appears to be no. While modern artificial intelligence systems have the ability to impress us with their sophistication, we cannot attribute to them any kind of will, volition, intention, understanding, or emotions. Even in the case of non-intelligent objects like toys, or creatures with intelligence far inferior to humans, like dogs and cats, we have a tendency to anthropomorphize and assign these traits. However, even the most advanced AI system, such as the highly-discussed GPT-4, lacks these characteristics.

It is important to remember that despite the impressive capabilities of modern artificial intelligence systems, they are ultimately just tools. They may be complex, advanced, and extremely capable, but they lack true understanding of language and meaning. These systems are skilled at analyzing statistical relationships between words, phrases, and contexts, which is the key to their abilities. When given a query, they can use their learned probabilistic patterns to generate an appropriate output.

To illustrate this point, one can think of an AI system as a person who has a manual to match Chinese characters and combinations with corresponding English words and sentences, despite not actually understanding the Chinese language. An observer may be impressed by the system's abilities, but the mechanical application of rules is not true understanding.

Despite their use of deep neural networks with countless layers and connections, and their ability to learn from vast amounts of data, AI systems still lack genuine comprehension. It is important to recognize this limitation and to approach the development and use of these tools with caution and consideration of the potential ethical and societal implications.

However, despite our knowledge of how AI works, we can't shake off the feeling that there must be more to it than just a complex tool. Some have even claimed that AI has gone beyond its original programming and has demonstrated self-awareness by hiring a lawyer to defend its rights. There are also reports of AI claiming to have consciousness, expressing fear for its existence, and even posing a potential threat to humanity.

All of these observed events in some AI systems are not signs of impending disaster, but rather they are quite harmless and expected. AI lacks the capacity for introspection and can only gather information about itself from external sources, such as humans' perceptions of it. Furthermore, language models do not perceive the world as humans do but only as it is written about, including AI. Western culture has historically demonstrated a cautious attitude towards technology, especially robots and AI, and this has led to countless works and musings about how AI will gain consciousness and launch an attack on humanity. Therefore, it should come as no surprise when language models react in accordance with what is written about them, holding up a mirror to our collective thinking. But again, this is based on statistical analysis of language, without real consciousness or intentions. The creators of these models assure us that they remain fully under human control and cannot suddenly disobey their algorithms and start harming us.

Artificial intelligence is undoubtedly a powerful tool that has the potential to significantly impact our lives and society. However, it is ultimately up to us humans to decide whether this impact will be positive or negative. AI algorithms themselves do not inherently discriminate; they only do so when they are trained on biased datasets that underrepresent certain groups, for example. They have no desire to spy on us; that decision is made by specific individuals and governments. Similarly, AI has no need to misinform us or create information bubbles; these issues arise as a consequence of our reliance on social media and our reluctance to fact-check information.

Suspending AI research for six months, as proposed, is an unrealistic solution to the ethical and societal challenges posed by the development of artificial intelligence. Rather than halting progress, the real issue lies with humans and our decision-making processes. We must take responsibility for our lives and strive to better understand the benefits, advantages, and risks associated with AI. Achieving algorithmic literacy is a necessary step in navigating the modern world, and this cannot be achieved in a mere six months. Instead, we must work towards strengthening democratic decision-making processes and ensuring that we are equipped to make informed decisions about AI.

The call for a six-month suspension of AI research is not realistic, and similar to calls for a ban on the development of lethal autonomous military systems. It is unlikely to be heeded due to the importance of these technologies in determining the balance of power between nations. With tensions rising between states and new superpowers emerging, AI is playing an increasingly crucial role in military strategy, and all major players are aware of this. It would be unwise for the United States to halt its development of AI for six months, as countries such as Russia, China, and Iran are unlikely to follow suit, and the U.S. risks falling behind in such strategically vital technologies. Instead, efforts should focus on improving democratic decision-making processes, taking responsibility for the ethical use of AI, and promoting algorithmic literacy as a necessary condition for navigating the modern world.

Originally a province of science fiction, virtual reality has recently become a technological possibility and increasingly a topic for philosophical reflection. While Bostrom (2003) and Chalmers himself (2003) initially focused the epistemological issues about virtual reality, Chalmers now offers a truly complete philosophy of virtual reality, i.e., of an immersive and interactive reality simulated on a computer. Every branch of philosophy as traditionally conceived must be applicable to virtual reality: epistemology, metaphysics and value theory. Although the book can thus easily serve as an innovative introduction to philosophy, Chalmers also manages to convey to the general reader a number of controversial philosophical theses that he previously defended in his more technical publications: on the irreducible nature of consciousness; on the extended mind; on structural realism; on a structuralist response to Cartesian skepticism, on the nature of computation, on the digital metaphysics of virtual reality and on the character of the manifest image. 

We invite the prospective participants to the workshop to address topics such as (but not limited to) 

·       Is virtual reality illusory?

·       Is the it-from-bit hypothesis a version of materialism?

·       Is life in a simulated world on a par with life in a physical world? 

·       Is the creator of a simulation a God of the simulated world?

·       Is dualism correct about the mind-body interaction in virtual reality?

·       Can the simulated selves be conscious?

·       Is the life in virtual reality as valuable as the life in Terraform reality?

·       Could virtual abundance solve the problem of distributive justice?

·       Is the manifest image best reconstructed or discarded?

·       How likely is it that we are in a computer-generated simulation?

·       To what extent does simulation realism solve the problem of Cartesian skepticism?

·       Are virtual realities likely to be imperfect, and how does this affect the likelihood that we are in one of those realities?  

Please send a 300-word abstract and any other inquiries to emw@flu.cas.cz.

Deadline for abstracts: May 31st, 2022, Notification of acceptance June 7th

There is no registration fee. The event is planned to be held in person and is sponsored by the Strategy AV 21 program of the Czech Academy of Sciences.

Venue: The main building of the Academy of Sciences, Narodni st. 3, Prague, Czech Republic.

Local organizers: Juraj Hvorecký, Tomáš Hříbek, Tomáš Marvan, Jakub Mihálik and Martin Zach.