Consciousness Credit:uxmag.com
Introduction
Whether AI can be conscious is the subject of intense, unresolved debate among scientists and philosophers. While AI is currently not conscious, experts say there are no fundamental, theoretical barriers to it developing forms of awareness as systems become more advanced and complex. Most experts believe that AI systems based on Large Language Models (LLMs) are sophisticated calculators – processing data and predicting text – rather than being sentient or encompassing feelings. Yet some believe that AI could potentially achieve some form of consciousness. However, many argue that it would require biological components like a human body, making artificial consciousness impossible. In the case of AI achieving consciousness, some researchers worry about risks and ethical issues. Nonetheless, many suggest that new models of information processing might eventually meet the criteria for consciousness, raising both profound scientific questions and significant ethical challenges.
From Science Fiction to Reality
The idea of AI machines with their own minds has long been explored in science fiction. A fear of machines becoming conscious and posing a threat to humans was explored in the 1968 film 2001: A Space Odyssey, when the HAL 9000 computer attacks astronauts onboard their spaceship to take control. Recently credible researchers have become concerned that this is no longer an issue of science fiction. As AI becomes more intelligent, AI machines will become conscious as well. Others, such as Prof. Anil Seth from the University of Sussex , disagree, describing the view as blindly optimistic. Consciousness is associated with intelligence and language because they are interlinked. At the same time there is no approved definition of consciousness. Some researchers from the tech sector believe that the AI in our computers and phones may already be conscious and we should treat them accordingly. Google suspended software engineer Blake Lemoine in 2022, after he argued that AI chatbots could feel things and are sentient. According to Prof. Seth we may well be living in a world populated by humanoid robots and deep fakes that seem conscious. But he worries that we will not be able to resist believing that AI has feelings and empathy. This means that we trust these AI-systems more, share data with them and are more open to persuasion. As a result human relationships will be replicated in AI relationships, and these systems will be used as teachers, friends, adversaries in computer games and even romantic partners. Whether that is a good or bad thing is difficult to predict, but it is going to happen, and we will not be able to prevent it.
From Sentience to Consciousness
Sentience implies a basic form of consciousness that includes feelings and sensory perception. A fundamental challenge in studying sentience is that subjective experiences like pain cannot be directly measured, even in humans who can self-report their pain and describe its severity. The Cambridge Declaration on Consciousness established a scientific consensus that humans are not the only sentient beings but possess neurological structures complex enough to support conscious experiences. The concept of individual sentience profiles recognizes that each individual experiences events or environments in unique ways and assigns specific values to those experiences. This approach highlights inter- and intra-species variability in sentient experience and value assignment, suggesting that sentience does not manifest identically across or within species. Individual sentience profiles can accommodate individual differences in behavioural and physiological responses to experimental challenges and caretaking procedures. The assessment and consideration of individual sentience profiles in neuroscience research is proposed as a means to improve welfare and refine research outcomes, though practical implementation requires balancing welfare considerations with experimental control.
The Neuroscience View
To well-known researcher Christoph Koch, the surge of people attributing consciousness to their chatbots is bizarre. The trend erodes complex human relationships and in his view massively devalues the human experience. He wonders why conscious creatures like us are willing to devote our lives to ‘something unconscious,’ surrendering ground to sophisticated yet lifeless mechanisms. The following lists some key points of his assessment:
Consciousness is not about doing: We reward doing far more readily than we value being, or experience. In our capitalist societies we value work that relates to intelligence, whether physical or intellectual. What matters is not what you think, dream or imagine, it is what you do. In a world where behaviour carries the greatest weight, machines that perform the same tasks we perform start to resemble us in unsettling ways.
How consciousness got cancelled: Conscious experience, what Koch calls ‘the feeling of life itself’ should address the most important issues of consciousness. Yet, our culture celebrates intelligence while conscious experience struggles to be recognized. Until recently, Koch notes that textbooks of psychology routinely left out the description of conscious experience. Even neuroscience texts rarely question what it feels like to be the owner of a brain. Such questions were considered beyond the reach of science. Some philosophers still consider inner life as an illusion, making us believe that our experiences are fake. Likewise ‘Free Will’, is considered an illusion as well, stripping away the very qualities that distinguish us from machines.
Future outlook
Koch is convinced that consciousness cannot be disregarded forever. One scientific form of consciousness is defined by the ‘Integrated Information Theory’. Viewed through this lens, consciousness is not a computation, and not an input-output process or a function. It is a structure grounded in the physics of complex systems. The theory measures consciousness through integrated information and the degree to which a system forms an irreducible whole. The more integrated a system is, the more it experiences the outside world. AI-systems with high integration are capable of making decisions and digital computers may eventually outthink us, but that represents action without experience. Intelligence is computable, consciousness is not.
The Philosophical View
According to Wikipedia there are many hypothesized types of consciousness and there are many potential implementations of artificial consciousness. In the philosophical literature, perhaps the most common classification of consciousness distinguishes ‘access’ from ‘phenomenal’ variants. Access consciousness concerns those aspects of experience that can be apprehended, while phenomenal consciousness deals with those aspects of experience that cannot be apprehended. In 2022, Google engineer Blake Lemoine made a viral claim that Google’s LaMDA chatbot was sentient. He supplied as evidence the chatbot’s human-like answers to many of his questions. However, the chatbot’s behaviour was judged by many scientists as a consequence of mimicry, rather than machine sentience. Back in 2023 David Chalmers argued that LLMs display impressive conversational and general intelligence abilities, but are not yet conscious, as they lack some features that may be necessary, such as the inclusion of a global workspace. Nonetheless, he considers that non-biological systems can be conscious and suggests that future large language models incorporating these elements, might eventually meet the criteria for consciousness. If an intelligent machine is conscious, its rights would need to be assessed as well. Because artificial consciousness is still largely a theoretical subject, the ethics related to its rights have not been discussed or developed although it has often been a theme in science fiction. In 2021, German philosopher Thomas Metzinger argued for a global moratorium on artificial consciousness until 2050. Metzinger asserts that humans have a duty of care towards any sentient AIs being created and that proceeding too fast an ‘explosion of artificial suffering’ might occur. Likewise David Chalmers argues that by creating conscious AI would raise difficult ethical challenges, with the potential for new forms of injustice.
Conclusion
Among the research community there seems to be agreement that AI-systems have or will achieve some form of sentience and consciousness. Some warn about the ethical risks and the legal issues that need to be addressed regarding the rights conscious machines could demand. Due to the increasing pace of technology and its application for solving problems humanity might lose control over its consequences. In contrast, human consciousness differs widely from AI-systems. The human body and its sensory inventory cannot be replaced by intelligent machines. Humans have feelings and empathy, something an intelligent machine clearly lacks. Regardless of the neuroscientific or philosophical view there are many unresolved issues. New Large Language Models (LLMs) might provide an indication of AI’s future contribution to artificial consciousness. However, the research of human consciousness and the methods available are still in its infancy. Far more research capacity and funding is necessary to provide a better understanding about future developments.