Have you ever engaged in a conversation with someone who is deeply interested in consciousness? How did that conversation unfold? Did they make vague hand gestures in the air or reference complex philosophical works? Perhaps they shared their belief that reality is subjective and that scientists cannot truly understand it. The study of consciousness has traditionally been disregarded by the natural sciences due to its elusive nature. It has mainly been left to philosophers who often struggle to provide clear explanations. In fact, consciousness has been mockingly referred to as “the C-word” by some in the field of robotics.
However, a recent report by a group of philosophers, neuroscientists, and computer scientists, including Dr. Grace Lindsay from New York University, proposed a framework for determining whether an artificial intelligence (AI) system such as ChatGPT could possess consciousness. This groundbreaking report, which explores the emerging field of consciousness science, combines insights from several empirical theories and suggests a set of measurable qualities that could indicate the presence of consciousness in a machine.
One theory discussed in the report is recurrent processing theory, which focuses on the distinction between conscious and unconscious perception. Neuroscientists argue that unconscious perception occurs when electrical signals pass through the nerves in our eyes and primary visual cortex, while conscious perception occurs when these signals are then passed back to the primary visual cortex from deeper parts of the brain, creating a loop of activity.
Another theory proposes the existence of specialized brain sections dedicated to specific tasks. For example, the part of the brain responsible for balancing on a pogo stick is different from the part that appreciates a beautiful landscape. Neuroscientists believe that a “global workspace” coordinates these separate brain sections, allowing us to integrate and control our attention, memory, and perception. This integrated workspace could be the source of consciousness.
Alternatively, consciousness might arise from the ability to be aware of one’s own awareness, create mental models of the world, predict future experiences, and navigate one’s physical body. The report suggests that any of these features could be crucial components of consciousness. If these traits can be identified in a machine, then that machine may be considered conscious.
One challenge with this approach is that advanced AI systems, like deep neural networks, learn independently in ways that are not easily interpretable by humans. The “black box problem” of AI arises because we can only extract limited information from their internal structures. Therefore, even with a comprehensive rubric of consciousness, it would be challenging to apply it to the AI systems we encounter daily.
The authors of the report emphasize that their list of qualities for consciousness is not definitive. They rely on a concept called “computational functionalism,” which reduces consciousness to information exchanged within a system, similar to a pinball machine. According to this view, a pinball machine could potentially be conscious if it became significantly more complex. However, other theories propose that consciousness involves biologically or physically ingrained characteristics and relies on social and cultural contexts, which would be difficult to code into a machine.
Even among researchers who align with computational functionalism, there is no agreement on a theory that fully explains consciousness. Dr. Lindsay acknowledges that the report’s conclusions depend on the accuracy of the theories, and currently, they fall short. This uncertainty highlights the elusive nature of consciousness.
Ultimately, can any combination of these features truly capture the richness of conscious experience described by William James as “warmth” or Thomas Nagel as “what it is like” to be oneself? There remains a gap between the subjective experience of consciousness and our ability to measure it scientifically. This gap represents the “hard problem” of consciousness, as described by David Chalmers. Even if an AI system possesses recurrent processing, a global workspace, and self-awareness, what if it lacks the subjective feeling of consciousness?
When I discussed this dilemma with Robert Long, a philosopher leading the report at the Center for A.I. Safety, he noted that struggling to explain high-level concepts through scientific or physical processes often leads to a sense of emptiness. This feeling arises when we attempt to scientifically explain a phenomenon that transcends purely mechanical processes.
The stakes are high, as AI and machine learning continue to advance at a rapid pace. In 2022, Google engineer Blake Lemoine claimed that their chatbot, LaMDA, was conscious (although most experts disagreed). As generative AI becomes more integrated into our lives, the debate over machine consciousness is likely to intensify. Dr. Long argues that we must begin taking a stand on what machines might possess consciousness and criticizes the vague and sensationalist approach often taken. He believes this is a pressing issue that we must address in the coming years.
As Megan Peters, a neuroscientist and report author, points out, whether or not consciousness exists in an entity greatly influences how we treat it. We already conduct similar research with animals to determine if they have experiences akin to our own. This process resembles navigating a funhouse, shooting arrows at shape-shifting targets from moving platforms, with the occasional realization that our bows are made of spaghetti. However, sometimes we do hit the mark. As Peter Godfrey-Smith wrote in “Metazoa,” cephalopods likely have a different but equally vivid subjective experience compared to humans. Octopuses have around 40 million neurons in each arm. What must that be like?
To unravel the mystery of consciousness, we employ a range of observations, inferences, and experiments, both controlled and spontaneous. We communicate, interact, play, hypothesize, prod, analyze, and dissect. Yet, ultimately, we still lack a definitive understanding of what makes us conscious. We can only assert that we are conscious beings.
Denial of responsibility! Vigour Times is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
