Approaches on whether AI should have consciousness. PART I

Through this work, based on the research, notes, ideas, and projects that my team and I have undertaken, I aim to reveal the intricacies involved in creating an AI and integrating it into work, production, or service processes. The process of adoption extends well beyond simply connecting to an API that can solve a particular problem. Instead, this work emphasizes the need to comprehend how an AI should truly think by understanding the functioning of our brain and consciousness. Without delving into the theories that have been attempting to uncover the workings of our brain for millennia, it is impossible to assert that we are working on an AI project, let alone an AI project that is genuinely practical and, most importantly, ethical.

It’s worth noting that this work is extensive, so I plan to release it in smaller parts. My hope is that these parts can prove useful to those who, beyond the computational aspect, need to understand why they should use AI-based technologies and how to approach the design process.

Introducing the theory of Phenomenal Consciousness

Is it possible to create a programmed machine with the ability to have a subjective consciousness? Could a human-created and algorithm-based artifact experience subjective things like color, pain, scent, enjoy music, or feel stress?

Since the inception of Artificial Intelligence (AI) with Alan Turing’s seminal work in 1948 and 1950, the discipline has sought to create machines capable of emulating and competing with human cognitive abilities using algorithms as a cornerstone. An algorithm is an ordered series of finite steps that lead to a result, and it is what powers the so-called “Turing Machine”. The basic idea of this classical approach in AI is that, if a machine is equipped with the right algorithm, it should be able to perform any task a human can do.

Due to this “optimism,” the discipline has been interested in creating machines capable of emulating other characteristics of human thought, with special attention paid to the development of machines that exhibit consciousness. This is why in the last decade, there has been a growing interest in what is known as “Artificial Consciousness” (also known as “Machine Consciousness” and “Synthetic Consciousness”), which seeks to design and implement models of artificial consciousness. The reasons for this enterprise can be grouped around four central motivations: first, to design consciousness models to determine their defining traits in relation to work in cognitive science; second, to use the implementations of these models as tools to test and understand human consciousness; third, to determine how and what is the relationship between consciousness and the physical support on which it occurs; and finally, to understand that phenomenal consciousness is fundamental in some of the tasks considered intelligent.

Literature and cinema have explored the concept of advanced machines capable of complex tasks, such as HAL 9000 in ‘2001: A Space Odyssey’, robots in Isaac Asimov’s ‘I, Robot’, or conscious machines in Steven Spielberg’s ‘AI: Artificial Intelligence’. These works raise the idea of machines that have consciousness and abilities similar to humans. Real-life examples also exist, such as virtual assistants like Siri and Alexa that can learn and respond to users’ voice commands.

Creating machines that mimic human characteristics, such as adding an appropriate part, is still a debated issue in Artificial Intelligence. While the discipline has advanced in creating machines with cognitive capabilities similar or superior to humans, there is still debate about whether these machines can exhibit a phenomenological consciousness similar to humans. The ability to have subjective experiences, such as tasting tea or enjoying music, remains a topic of discussion in AI.

Additionally, questions are raised about how to determine if a machine possesses such consciousness, the relationship between physical support and its appearance, and whether machines can think or simply perform calculations. Although machines like Deep Blue have demonstrated outstanding abilities in games like chess, it is still questioned whether they really understand the game or simply follow a set of predetermined steps.

The subjective experience of playing chess is something that Deep Blue does not possess, as there is nothing in its design or algorithms that allow it to have such an experience or any other experience related to the game, such as the tension or pressure of competition or the influence of the opponent’s moves. Supporters of the so-called “Strong AI” would argue that it is only necessary to observe the external behavior of the machine to determine if it possesses traits of intelligence or intelligent activity. This is illustrated in the “Turing Test,” where a machine is considered conscious if it can pass tests that humans have established as sufficient to demonstrate consciousness. In fact, this is what we do all the time with other human beings, assuming that they also have consciousness due to how they behave and appear to us.

This raises the point that the behaviorist claim that the existence of consciousness in a machine can be determined by objective tests is not applicable to machines, as consciousness is something that can only be subjectively experienced by the possessor of that consciousness. Additionally, it is pointed out that this problem is not exclusive to artificial intelligence but is also a problem in the philosophy of mind, which has given rise to different approaches such as dualism and monism. Although the discussion about the mind-body problem has existed for several centuries, there is still no consensus today, and the diversity of sub-arguments has contributed to a lack of unanimity on this topic.

Neurobiologist Bernard Baars highlights the confusion surrounding the topic of consciousness, which stems from the classic mind-body paradox. This paradox arises when questions about consciousness are posed in terms of subjectivity, or what it feels like to be someone. This can lead to one of the three classic positions on the problem: mentalism, physicalism, or dualism.

Baars suggests that in order to answer questions about consciousness, it is necessary to frame them in a way that can be answered, using language and an approach that allows for a more concrete and less abstract way of addressing the problem.

However, there is complexity and a lack of consensus regarding the definition and aspects of the term “consciousness.” Philosopher of mind Ned Block argues that consciousness is a “hybrid concept,” meaning that it refers to several different phenomena related to the subjective experience of the world, such as attention, awareness of something external or internal, individual identity, the ability to experience the world from a subjective point of view, and experiential phenomena like color and pain.

The text also highlights that there are diverse philosophical approaches to the nature of the mind and consciousness, and within each of these approaches there are various positions and arguments. This lack of consensus and complexity makes it difficult to answer questions about consciousness and suggests the need for more precise definitions and well-formulated questions.

In addition, the views of David Chalmers, particularly his defense of the “hard problem” or “philosophical zombie” argument, should also be taken into account. He argues that while we could imagine a world in which everything behaves identically to our current world but where there is no consciousness, the mere possibility of this scenario suggests that consciousness cannot be reduced to purely physical or material explanations. Chalmers has extensively written about the nature of consciousness and proposed various theories, including the notion of “panpsychism,” which holds that consciousness is a fundamental property of the universe as a whole.

Throughout this work, you will see a critical and constructive analysis of the possibility of a machine programmed with algorithms having phenomenal consciousness, i.e., the subjective experience of the world that a being experiences when interacting with its environment and inner life. Although there are other notions of consciousness, this analysis focuses specifically on this ability, without delving into the possibility of the machine exhibiting other types of consciousness, such as self-consciousness or access consciousness. This approach is relevant to better understand what consciousness is and to analyze its key components before discussing the possibility of modeling the subjective experience of the world in a machine.

It is important to discuss the philosophical foundations that support the work of disciplines such as AI and psychology. Additionally, the discussion about consciousness in programmed machines raises ethical and legal questions that should be considered. Finally, it is essential to highlight the importance of endowing programmed machines with phenomenal consciousness, as this ability is fundamental to certain tasks considered intelligent, unlike approaches that consider it an irrelevant epiphenomenon.

While it is argued that achieving this task is possible and necessary, there is a question of whether modeling through algorithms is the best alternative to achieve it, as algorithms do not explain how or why this particular trait of cognitive activity occurs. One alternative approach is to explore the possibility of constructing an artificial neural network that would simulate the biological neural network of the brain, which could provide a more accurate model for consciousness. However, this approach also presents significant challenges, such as the difficulty of reproducing the complexity of the brain and the issue of determining how to evaluate the presence of consciousness in such a network.

Overall, this work aims to contribute to the ongoing discourse about consciousness in programmed machines, highlighting the complexity of the issue and the importance of thoughtful and deliberate consideration of the implications and possibilities of this field.

Continuará…

#WebScraping #artificialintelligence #bigdata #datascraping #prompt #datamining #inteligenciaartificial #innovation #technology #futurism #digitalmarketing #GenAI #AI #IA #fakenews